Opinion
18 C 8175
08-20-2024
KOVE IO, INC., Plaintiff, v. AMAZON WEB SERVICES, INC., Defendant.
MEMORANDUM OPINION AND ORDER
MATTHEW F. KENNELLY United States District Judge
Kove IO, Inc. (Kove) brought this lawsuit in 2018 against Amazon Web Services, Inc. (AWS) for infringing three of Kove's patents. In response, AWS asserted various affirmative defenses and counterclaims. The Court granted in part and denied in part both parties' motions for summary judgment. See Kove IO, Inc. v. Amazon Web Servs., Inc., No. 18 C 8175, 2024 WL 450028 (N.D. Ill. Feb. 6, 2024).
In April 2024, the case was tried before a jury, which found in favor of Kove on its patent infringement claims but declined to find willful infringement. The jury awarded Kove $525,000,000 in lost royalty damages. Both parties challenge the judgment. AWS has moved under Federal Rule of Civil Procedure 50(b) for judgment as a matter of law (JMOL) on infringement and damages. In the alternative, AWS has moved under Rule 59(a) for a new trial based on particular rulings by the Court and on the question of damages.Kove has moved to amend the judgment to include pre-judgment and post- judgment interest. Kove has also moved for attorneys' fees under 35 U.S.C. § 285. The Court addresses each request in turn.
For patent cases, the standards for motions for judgment as a matter of law and motions for a new trial are determined by regional circuit law. Finisar Corp. v. DirecTV Group, 523 F.3d 1323, 1328 (Fed. Cir. 2008).
Background
A. Kove's patents
Kove is a Delaware corporation with its principal place of business in Chicago, Illinois. Kove is the owner of the three patents-in-suit: U.S. Patent Nos. 7,103,640 ('640 Patent); 7,814,170 ('170 Patent); and 7,233,978 ('978 Patent). The patents generally relate to the storage, search, and retrieval of data across large computer networks.
The '170 patent is a continuation of the '640 patent. Both the '640 and the '170 patents claim systems for "enabl[ing] hyper-scalable cloud storage and improv[ing] upon the scalability limitations of conventional storage systems." Compl. ¶ 18. The patents are directed towards addressing various constraints that limit the speed, capacity, and efficiency of conventional data retrieval systems. For example, distributed data collection systems store and retrieve data from multiple machines connected through a single network. At the time of invention, distributed data systems were generally arranged in a hierarchical configuration that included indices that required periodic updates. As the network became larger and more complex, these hierarchical indices became increasingly inefficient. Additionally, servers within these data collection systems typically were unable to find and retrieve data located in other servers throughout the network. The inventions claimed in the patents-in-suit provide a method for overcoming these constraints.
The preferred embodiment of the inventions "provides a network distributed tracking wire transfer protocol in a networked environment." '640 pat., col. 2, lns. 12-14; '170 pat., col. 2, lns. 14-16. A "client" queries one or more location servers in the network distributed tracking protocol (NDTP) server constellation for information about the location of data. An NDTP server responds to the query with location information, and the client uses that information to retrieve the desired data. If a particular location server does not contain the desired location information, the client sends a request to a location server, and the location server will respond with a redirect message that identifies the location server that contains the relevant information. This network distributed tracking wire transfer protocol also includes identification strings specifying the identity of the data and location strings specifying the location of the data within the network. The patents disclose a process by which the relationships between identification and location information are able to "change dynamically, spontaneously, and efficiently." '640 pat., col. 2, lns. 23-24; '170 pat., col. 2, lns. 24-25.
The '978 patent is a continuation-in-part of the '640 patent and discloses "a system and method for storing and retrieving location information across a network." Compl., Ex. 5 at 1. The specification explains that a redirect mechanism "permit[s] the distribution of identifiers to location mappings across members of an NDTP server cluster" containing multiple location servers. '978 pat., col. 2, lns. 11-13. The '978 patent is directed to "scaling system capabilities in a manner sufficient to handle variable demand for resources." '978 pat., col. 2, lns. 12-16. In contrast to conventional storage systems, the claimed technology allows for the quick and accurate identification of location servers that contain the location information of a particular file across large data storage networks.
B. AWS's products
AWS is a subsidiary of Amazon that specializes in cloud-computing platforms and services. Kove has identified two cloud-based AWS products that it believes infringe on its patents: Amazon Simple Storage Service (S3) and DynamoDB (DDB). S3 provides storage for companies building Internet applications and the product currently serves over two million websites. Tr. at 1264:4-1265:4. The pieces of data stored within S3 are called "blobs." S3 contains three key parts: the web server, the keymap system, and the volume system. The web server receives initial requests for data in the form of put requests, for storing data, and get requests, for retrieving data. The volume system stores the data objects within S3 and contains two main components: R2D2s, which are data repositories, and the Blob Assembler, which is used to assemble the data. The keymap system stores information about the data and consists of keymap functional coordinators (KFCs) and brick managers (BMs). While BMs store the entire keymap, KFCs store a subset of the keymap for objects in a local cache. The below diagram, taken from an Amazon video presentation, shows the basic architecture of S3:
(Image Omitted).
DDB enables users to create database tables to store, manage, and retrieve data items. In DDB, data is stored in tables, and tables are divided into partitions. Data requests are received by the request router. A request router can fulfill a limited number of data requests using information stored in a local cache. If the request router does not contain the necessary information, then it communicates with multiple components to fulfill the request, including storage nodes and metadata nodes. The following Amazon diagram demonstrates the general structure of DDB and its components:
(Image Omitted).
C. Present lawsuit
Kove filed suit against AWS in December 2018, alleging willful infringement of multiple claims in the '170, '640, and '978 patents. In response, AWS asserted numerous counterclaims including invalidity and unenforceability. In April 2019, AWS filed a motion to dismiss arguing that the three patents were invalid under 35 U.S.C. § 101. The Court (by Judge Pallmeyer) denied AWS's motion to dismiss. See Kove IO, Inc. v. Amazon Web Servs., Inc., 448 F.Supp.3d 849 (N.D. Ill. 2020).
On November 19, 2021, Amazon filed requests with the United States Patent and Trademark Office (USPTO) to reexamine multiple claims of the patents-in-suit. The USPTO granted Amazon's requests and subsequently upheld the patentability of the '640 and '170 patents. The USPTO later issued a final rejection for claims 1 and 31 of the '978 patent. Amazon has filed additional reexamination requests, which are currently pending.
In December 2021, the Court (by Judge Pallmeyer) construed numerous terms
within the patents-in-suit. Prior to the claim construction hearing, the parties agreed to constructions of the following terms:
| Claim term | Construction |
| Location | an encoding that is a member of a set of associations with an identifier in a location server, and that specifies where data pertaining to the entity identified by the identifier is stored |
| Identifier/identifier string/identification string | a unique encoding that identifies an individual entity, and with which zero or more location strings are associated in a location server |
| Hash table | a data structure that stores values in a table, where values are stored and retrieved by applying a hash function to an input and using the function result as an index into the table |
| Claim term | Construction |
| Location information | '978 patent: one or more identifiers and their associated locations '170/'640 patent: information pertaining to one or more locations of data and/or the identities of one or more location servers |
| Location server | a network-attached component that maintains a set of identifier/location mappings that are modified or returned in response to location request messages from clients |
| Client | a network-attached component (which may be software or hardware) that initiates update or lookup of identifier/location mappings from a location server with location request messages |
| "based on a hash function used to organize the data location information across the plurality of data location servers ... based on the hash function applied to the identifier string" | the portion of the data location information included in a corresponding one of the data location servers is based on a hash function that maps identifier strings to one or more of the data location servers, and each one of the data location servers is configured to determine the at least one of the plurality of data location servers based on the hash function applied to the identifier string |
| "data pertaining to the entity" | data pertaining to a person or thing (real, digital, or abstract) distinct from that data |
AWS later moved for summary judgment on each of Kove's claims. Kove filed a cross-motion for summary judgment. The Court denied AWS's motion for summary judgment on the issues of non-infringement and willful infringement. See Kove IO, Inc., 2024 WL 450028, at *11-16. This Court also granted summary judgment in Kove's favor on AWS's defenses of patent ineligibility, equitable estoppel, and waiver, and denied Kove's request for summary judgment on AWS's remaining defenses of inequitable conduct and double patenting. Id. at *16-27. Prior to the final pretrial conference, the parties filed a joint stipulation to narrow the issues for trial by reducing the number of asserted claims and the number of bases for certain of AWS's invalidity defenses. Dkt. no. 832 at 1. AWS also agreed to drop its inequitable conduct and double patenting defenses. Id. Five days before opening arguments, AWS dropped its remaining invalidity defenses. On the first day of trial, Kove dropped additional claims. The remaining claims at issue at trial were claims 1 and 2 of the '170 patent, claim 18 of the '640 patent, and claims 10, 17 and 30 of the '978 patent. Claim 18 of the '640 patent and claim 10 of the '978 patent were asserted only as to DDB.
C. Evidence at trial
The case was tried before a jury in April 2024. The jury heard testimony from multiple fact and expert witnesses and was shown numerous exhibits.
1. Expert testimony on infringement
Kove's infringement expert, Dr. Michael Goodrich, opined that S3 and DDB infringe upon the Kove patents. To develop his opinions, Dr. Goodrich reviewed Amazon technical documents, product literature, deposition testimony, training videos, and source code. At trial, Dr. Goodrich testified that KFCs and BMs are location servers because they store location information for objects that are stored in R2D2s. He explained that KFCs are able to identify which other KFC stores a specific set of location data through a hash function. He further explained that a unique identifier is entered as an input into the consistent hash ring, and that the output identifies the correct KFC. Tr. at 558:1-9. He further stated that BMs use a "tree function" to find the BM that contains the requested data. Id. at 558:13-23. Dr. Goodrich also explained that iNodes are a "location" within the meaning of the asserted patents because they contain a Blindex, which in turn contains the volume ID and the blob ID for a particular object. According to Dr. Goodrich, "the volume ID tells you the volumes of where you're going to store that on the R2D2s, and then the blob ID is the ID for this particular blob of data that's being stored." Id. at 559:1-14. Dr. Goodrich also walked the jury through every asserted claim in the patents-in-suit and explained why he believes the accused products meet the requirements of those claims. See, e.g., Id. at 567:14-568:3.
Dr. Grama, AWS's infringement expert, testified that the accused products failed to satisfy various limitations of the asserted claims, including the "location server," "identifier," and "location" requirements. At trial, Dr. Grama provided his noninfringement opinion on each asserted claim of the patents-in-suit. See, e.g., Id. at 1511:5-1514:10. He testified that within DDB, a component called the "ALF," rather than the metadata nodes, identify which metadata node contains the requested item. Id. at 1468:2-14. He further testified that a partition key cannot be a unique encoding that identifies an individual entity because it is "a key for the entire partition." Id. at 1459:14-23. Turning to S3, Dr. Grama stated that an iNode is "an internal name or a label" rather than a location. Id. at 1483:12-19. Dr. Grama testified that KFCs "maintain a very, very small fraction of the total data" and therefore cannot satisfy the limitation that each identifier is associated with at least one location. Id. at 1504:151505:10. He also disputed Dr. Goodrich's description of the consistent hash ring and stated that iNodes are not assigned to KFCs based on a hash function. He explained that KFCs "figure out what node is being accessed" and if the node is not available in the cache, "it goes to the brick manager . . . brings it back to the KFC, and keeps it there." Id. at 1499:18-1500:7. Dr. Grama concluded that neither S3 nor DDB infringe upon the asserted patents.
2. Expert testimony on damages
Kove described the benefits of distributed storage products using the acronym "SADS," which stands for "speed, availability, durability, and scalability." Id. at 430:1-3. Karim Fanous, one of Kove's expert witnesses, testified that these benefits are important to distributed database users and that customers are willing to pay more to access these benefits. Jim Bergman, Kove's damages expert, testified that to perform his damages analysis he consulted with Dr. Goodrich and Fanous and reviewed expert reports, deposition testimony, financial information, and industry research. He stated that for his damages analysis he calculated the potential cost of a royalty, which is "a payment for the right to use somebody's property." Id. at 1012: 20-1013:7. He explained that a royalty can be structured as a lump-sum royalty, which is paid upfront, or a running royalty, where payments are made over time based on the frequency of use. Id. at 1013:8-12. Bergman stated that he used a running royalty structure for his analysis because it has a better chance of "capturing the appropriate amount of use and the value of that use." Id. 1013:24-1014:12. He opined that, based on his analysis, in a hypothetical negotiation between the parties AWS would have agreed to pay between 50% and 100% of its quantifiable benefits from use of the patents for a license agreement. Id. at 1104:5-18. He also opined that a reasonable royalty payment to Kove falls between $517 million and $1.034 billion, Id. at 1010:7-13, and a reasonable lump sum payment would be between $320 million and $700 million. Id. at 1911:11912:1.
AWS's damages expert, Alyssa Bennis, also estimated the amount that the parties would have agreed on in a hypothetical negotiation for a license agreement to practice Kove's patents. Bennis opined that AWS would have agreed to a lump sum, up-front payment of no more than $5.25 million. Id. at 1643:10-16. After offering her opinion on the damages amount, Bennis testified that Bergman's analysis focused too heavily on Amazon's alleged use of the patents and that his damages calculations were "wildly overstated." Id. at 1645:21-1646:4; 1675:5-13.
Discussion
A. AWS's motion for judgment as a matter of law
AWS has moved for judgment as a matter of law on all of Kove's claims. Under Federal Rule of Civil Procedure 50(b), judgment as a matter of law is appropriate if the Court "finds that a reasonable jury would not have a legally sufficient evidentiary basis" to support a verdict for the nonmovant. FED. R. CIV. P. 50(a)(1). To determine whether the jury verdict had a legally sufficient evidentiary basis, the court must "construe the evidence strictly in favor of the party who prevailed before the jury, Passananti v. Cook County, 689 F.3d 655, 659 (7th Cir. 2012), and "draw[] all reasonable inferences in the light most favorable to [the non-moving party]." Kossman v. Ne. Ill. Reg'l Commuter R.R. Corp., 211 F.3d 1031, 1036 (7th Cir. 2000). The court also "disregard[s] all evidence favorable to the moving party that the jury [was] not required to believe." Passananti, 689 F.3d at 659 (quoting Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 151 (2000)). In reviewing the record, a court must neither "make credibility determinations" nor "weigh the evidence." Passananti, 689 F.3d at 659. Rather, a court must simply "determine whether a reasonable jury could have found in favor of" the prevailing party. Harvey v. Off. of Banks & Real Est., 377 F.3d 698, 707 (7th Cir. 2004).
"Overturning a jury verdict is a hard row to hoe." Zelinski v. Columbia 300, Inc., 335 F.3d 633, 638 (7th Cir. 2003) (citation omitted). Furthermore, "[i]nfringement is a question of fact." i4i Ltd. P'ship v. Microsoft Corp., 598 F.3d 831, 849 (Fed. Cir. 2010). "[T]he task of determining whether the construed claim reads on the accused product is for the finder of fact." PPG Indus. v. Guardian Indus. Corp., 156 F.3d 1351, 1355 (Fed. Cir. 1998).
In its Rule 50(b) motion, AWS addresses three issues. First, AWS argues that there was insufficient evidence that the accused products satisfy various claim limitations of the patents-in-suit. Second, AWS challenges the testimony of one of Kove's infringement experts, Dr. Goodrich. Finally, AWS asks the Court to vacate the jury's damages award.
1. Location/location server/identifier limitations
Each of the patents-in-suit contains system claims that require a "location," "location server," and "identifier." AWS contends that Kove failed to present evidence at trial that would allow a reasonable jury to conclude that the accused products satisfy these limitations. "[T]he failure to meet a single limitation is sufficient to negate infringement of the claim." Laitram Corp. v. Rexnord, Inc., 939 F.2d 1533, 1535 (Fed. Cir. 1991).
Prior to claim construction, the parties agreed that "location" means "an encoding that is a member of a set of associations with an identifier in a location server, and that specifies where data pertaining to the entity identified by the identifier is stored." Kove IO, Inc., 2021 WL 5988928, at *7. They also agreed that the term "identifier" means "a unique encoding that identifies an individual entity, and with which zero or more location strings are associated in a location server." Id. Judge Pallmeyer ruled that the proper construction of the term "location server" is "a network-attached component that maintains a set of identifier/location mappings that are modified or returned in response to location request messages from clients." Id. at 11-12.
AWS argues that S3 does not contain a "location" as defined by the claim construction, and therefore KFCs and BMs cannot be location servers. At trial, Dr. Goodrich testified that within S3, there are data structures called "iNodes" and that these iNodes contain a component, the "Blindex," that has a volume ID and a blob Id. Tr. at 559:1-14. He explained that "the volume ID tells you the volumes of where you're going to store [data] on the R2D2s, and then the blob ID is the ID for this particular blob of data that's being stored." Id. Dr. Goodrich opined that the combination of the volume ID and blob ID "specifies where a data object is stored," and therefore is a "location." Id. at 583:20-584:2. The jury viewed Amazon materials that refer to iNodes as "locations" or indicate that the data within an iNode is used to "tell S3 where the object is stored." See Id. at 582:14-583:19; 585:22-586:4. For example, Kove presented an excerpt of a video that features James Sorenson, an Amazon engineer, describing an iNode as a component that demonstrates "where your object is stored." Id. at 583:1219.
AWS contends that Dr. Goodrich's testimony was "conclusory." Def.'s Rule 50(b) Mot. at 2. On the contrary, Dr. Goodrich explained his conclusions at length, and he supplemented his opinion with S3 source code, deposition testimony, and various Amazon documents. See Tr. at 559:15-23; 584:21-585:21; 587:18-22. AWS does not identify any specific deficiencies with Dr. Goodrich's testimony and simply notes that AWS's witnesses with "first-hand knowledge" testified that a Blindex is "analogous to a label" and therefore is not a location. Def.'s Rule 50(b) Mot. at 2. AWS presented this argument to the jury, see Tr. at 1292:22-1293:1, and the jury disagreed. The existence of competing testimony is insufficient to overturn the jury's verdict unless the evidence "make[s] only one finding on the point reasonable." MobileMedia Ideas LLC v. Apple Inc., 780 F.3d 1159, 1168 (Fed. Cir. 2015). That is not the case here. Goodrich opined that the combination of the volume ID and the blob ID within the Blindex "specifies where the entity identified by the identifier is stored." Tr. at 584:21-585:21. The jury also viewed an Amazon wiki page that stated that the Blindex "tells us here the data is stored for a given object within the Storage layer." Id. at 585:22-586:6; PTX 699. Given Kove's competing evidence, it was not unreasonable for the jury to conclude that a Blindex is a location.
Additionally, "[c]redibility determinations . . . are jury functions, not those of a judge." Reeves, 530 U.S. at 150. AWS provides no legal authority to support its proposition that a jury's verdict must be consistent with the testimony provided by individuals with "first-hand" knowledge of an accused product in order to be reasonable or sustainable. And "[w]here there is substantial evidence for a reasonable jury finding, it is not our function to second guess or reevaluate the weight given to that evidence." MobileMedia Ideas LLC, 780 F.3d at 1168. The jury heard sufficient evidence from which it reasonably could conclude that S3 satisfies the location and location server requirements of all asserted claims.
AWS also asserts that DDB has "no location and no unique identifier mapped to a location in the Metadata Nodes" and therefore does not qualify as a location server. Def.'s Rule 50(b) Mot. at 3. As discussed above, the Court (by Judge Pallmeyer) construed "location server" to require "a network-attached component that maintains a set of identifier/location mappings." Kove IO, Inc., 2021 WL 5988928, at *11-12. It is undisputed that there are two kinds of primary keys within DDB: a simple primary key consisting of a single partition key and a composite primary key that contains a partition key and a sort key. Dr. Goodrich opined that these "primary keys" serve as identifiers because they uniquely identify items within tables in DDB. Tr. at 638:13-639:8. Dr. Goodrich also opined that the storage node IDs within DDB meet the "location" requirement. Dr. Goodrich explained that AWS engineers left comments on the DDB source code explaining how the technology works. The jury viewed one comment stating that the "node endpoint is the logical ID to uniquely identify an endpoint." Id. at 653:8-24. Dr. Goodrich explained that the endpoint "encapsulates both the logical ID and the physical endpoint address." Id. Additionally, the jury viewed an excerpt of deposition testimony from Sorenson, who identified "node IDs'" as a "piece of information" that is related to the location. Id. at 652:24-653:7.
Dr. Goodrich testified that identifiers are mapped to locations within DDB because the metadata nodes "store this mapping that starts from the primary key, which then is hashed . . . into the partition ID, which then is mapped to the storage node." Id. at 1776:16-24. AWS objects to this testimony as "evasive." Def.'s Rule 50(b) Reply at 3. But this was not the only point in Dr. Goodrich's testimony where he discussed the relationship between primary keys and storage node IDs. See Tr. at 650:25-651:17. He explained that a partition key, which is contained in both simple and compound primary keys, is put into a hash function that "organizes the way that data is spread across both the storage nodes and the metadata nodes." Id. at 666:10-23. Dr. Goodrich also walked the jury through a portion of DDB source code that he said demonstrates the process by which the hash function is used to determine the item's partition ID, and the partition ID is then used to locate the correct storage node ID or metadata node. Id. at 667:8-668:7.
AWS asserts that Kove is conflating primary keys and partition keys. But Dr. Goodrich clearly explained to the jury the difference between primary keys and partition keys. Tr. at 635:11-636:21. He also explained that his opinion concerning the mappings between primary keys and storage node IDs was based in part on the presence of partition keys within primary keys. Id. at 854:4-15. On cross-examination, when asked if "metadata nodes contain a mapping of partition IDs to storage node IDs," Dr. Grama responded, "[t]hat is correct." Id. at 1577:22-25. Given the jury's verdict, this Court infers that the jury found Dr. Goodrich's testimony on how this limitation is met through the presence of partition keys in primary keys persuasive and credible, as it was entitled to do. In its reply brief, AWS also contends that the slides Dr. Goodrich used to support his testimony regarding primary keys and partition keys within DDB were "misleading." Def.'s Rule 50(b) Mot. Reply at 2. But "the jury was assessing the weight and credibility of Dr. [Goodrich]'s expert testimony," not his slides. Aqua Connect, Inc. v. TeamViewer US, Inc., No. 181572, 2023 WL 6387791, at *4 (D. Del. Sept. 29, 2023) (noting that expert's demonstrative slides were not admitted into evidence or provided to the jury). It was not unreasonable for the jury to credit Dr. Goodrich's opinion that DDB met the "identifier," "location," and "mapping" requirements.
AWS also suggests that to maintain mappings of primary keys to storage node IDs, metadata nodes must "use the primary key." Def.'s Rule 50(b) Reply at 4. But there is no requirement in the claim limitation that the location servers must "use" the identifier competent. To the extent AWS wishes to clarify the definition of the word "maintain" as used in the Kove patents, this suggestion is untimely because AWS did not seek claim construction on this word. The Court never construed the word "maintain" to mean "use." And "[w]hen a claim phrase is not construed, we defer to the jury's view of the claim element unless that view is contrary to the only reasonable view of the claim element." VLSI Tech. LLC v. Intel Corp., 87 F.4th 1332, 1341 (Fed. Cir. 2023). AWS's definition of "maintain" is not the only reasonable view of the claim language, and AWS cites no support for its proposition that the term "maintain" requires "use." See Merriam-Webster Online Dictionary, https://www.merriam-webster.com/dictionary/maintain (defining "maintain" as "to keep in an existing state (as of repair, efficiency, or validity): preserve from failure or decline.").
AWS's points out that its own witnesses disputed portions of Dr. Goodrich's testimony. But "[i]t was up to the jury to determine credibility and resolve conflicts in the evidence." Wash World Inc. v. Belanger Inc., 666 F.Supp.3d 808, 822 (E.D. Wis. 2023). AWS had the opportunity to challenge Dr. Goodrich's testimony on this issue during cross-examination. See Tr. at 851:21-854:15. The jury heard AWS's experts' opinion on why the accused products do not satisfy this limitation, and it was entitled to determine that Kove's witnesses were more persuasive. Based on the trial record, the jury reasonably could determine that metadata nodes maintain mappings of primary keys to storage node IDs.
Finally, AWS attempts to adopt by reference over a dozen pages of arguments included in the briefs it submitted in support of its summary judgment motion. Def.'s Rule 50(b) Mot. at 3-4. "Adopting by reference arguments in documents other than in the brief dealing with the particular point under consideration not only provides an effective means of circumventing the page limitations on briefs . . . but unnecessarily complicates the task of the . . . judge." Miller UK Ltd. v. Caterpillar, Inc., 292 F.R.D. 590, 592-93 (N.D. Ill. 2013) (cleaned up). This Court deems any arguments incorporated by reference waived. See Hudgins v. Total Quality Logistics, LLC, No. 16 C 7331, 2024 WL 1618363, at *6 (N.D. Ill. Apr. 15, 2024) (Kennelly, J.) (rejecting party's attempt to incorporate arguments by reference).
For the foregoing reasons, AWS is not entitled to judgment as a matter of law on the location/location server/identifier limitations.
2. Non-hierarchical limitation
At summary judgment, this Court found that all asserted claims require a "plurality of location servers" in a "non-hierarchical" configuration. Kove IO, Inc., 2024 WL 450028, at *9. This Court adopted the definition of non-hierarchical that Kove presented to the USPTO during the reexamination process and concluded that the term means "any given server is able to return either the requested information or information usable by the client to locate the server with the requested information." Jury Instructions at 17. AWS argues that S3 and DDB cannot meet this limitation because the alleged location servers are configured hierarchically. AWS asserts that the jury heard testimony confirming the hierarchical configuration of the accused products from witnesses with "first-hand knowledge of the systems," but it does not provide examples of such testimony. See Def.'s Rule 50(b) Mot. at 4. AWS cannot meet its burden by "relying on the Court to sift through the trial record to determine what testimony (or lack thereof) supports [its] position." Brooks v. City of Chicago, No. 13 C 3090, 2018 WL 4404626, at *2 (N.D. Ill. Sept. 17, 2018).
Additionally, Dr. Goodrich testified that S3satisfies this claim limitation. He explained that within S3, KFCs are able to return an iNode (which contains a location) to a web server (the client) upon request if the iNode is stored in the local cache. Tr. at 571:23-572:4. He further testified that even if the location is not stored in its local cache, the KFC can still return the iNode by communicating with a BM. Id. at 572:24573:16. The jury viewed an excerpt of deposition testimony from Seth Markle, an AWS engineer, where he stated that when the KFC receives a key, it first checks to see if it has the key in its cache, and if it does, it "returns the iNode from its cache." Id. at 572:11-19. The jury also viewed a portion of S3 source code, which Dr. Goodrich stated demonstrated that KFCs "can access the consistent hasher to determine which KFC is assigned to a cache with any given iNode." Id. at 575:9-576:1. Dr. Goodrich also testified that the process by which a BM returns location information in response to a request from a KFC satisfies this claim limitation. Id. at 609:1-18.
AWS asserts that its argument applies to both S3 and DDB. Def.'s Rule 50(b) Mot. at 4 ("As AWS has explained, S3 and DDB are both configured hierarchically."). But it only addresses Dr. Goodrich's testimony about KFCs and BMs, which are components of S3. See id. at 5-6.
AWS provides three reasons why it contends Kove's evidence was legally insufficient, but none are persuasive. First, in support of its argument, AWS again attempts to incorporate arguments from its summary judgment briefing. Def.'s Rule 50(b) Mot. at 4 n.23. "[T]hese arguments incorporated by reference are inadequate and undeveloped and are therefore waived." United States v. Williams, 218 F.Supp.3d 730, 741 (N.D. Ill. 2016).
Second, AWS contends that Dr. Goodrich improperly "relied on the alleged location servers in a particular grouping or region as the required 'plurality of location servers.'" Def.'s Rule 50(b) Mot. at 5. But the portion of Dr. Goodrich's testimony that AWS cites contains no discussion of the "plurality of location servers" limitation. See Id. n. 24 (citing Tr. at 763-64). Moreover, at summary judgment this Court considered and rejected AWS's argument that it was improper for Kove's infringement contentions to rely on groupings of location servers within a specific zone or region. Kove IO, Inc., 2024 WL 450028, at *13 ("[A]s long as the location servers within an availability zone or region are arranged in a non-hierarchical configuration, the accused products may infringe upon Kove's patents even if those zones and/or regions are arranged hierarchically in relation to other components of S3 or DDB.").
Third, AWS asserts that Goodrich "admitted" at trial that the location servers he identified as being grouped in a plurality "wouldn't have all the information necessary to satisfy this Court's claim constructions." Def.'s Rule 50(a) Mot. at 5. But that is not a fair description of Dr. Goodrich's testimony. AWS first points to Dr. Goodrich's confirmation that SkyNet, a component of S3, "knows where every volume in the system is located." Tr. at 812:18-20. But in response to the very next question from AWS's counsel, Goodrich rejected the assertion that SkyNet, rather than the Blindex, specifies where data is stored within S3. Id. at 812:21-23. Dr. Goodrich further explained that because "the volume ID is sufficient to now say where the location is," that component of the Blindex along with the blob ID qualifies as a location. Id. at 813:2-7. Dr. Goodrich's direct rebuttal of AWS's contentions is far from an "admission" that the accused products do not meet the limitation at issue. Similarly, AWS cites to Dr. Goodrich's statement confirming that if a KFC does not have the requested iNode, it "goes to a Brick Manager." Id. at 841:2-4. But Dr. Goodrich repeatedly disputed the assertion that this process evidences a hierarchical configuration among KFCs and BMs. Dr. Goodrich testified that BMs and KFCs are connected though "communication paths" rather than different layers of a hierarchy. Id. at 841:10-18; 567:14-25. Although AWS's witnesses testified that the KFCs and BMs are arranged in a hierarchical manner, the jury was entitled to reject that evidence. AWS is not entitled to judgment as a matter of law based on the asserted claims' "non-hierarchical" requirement.
3. Other claim limitations
AWS argues that Kove failed to show that the accused products meet other limitations of the asserted claims. But in doing so, AWS largely ignores the standard of review for judgment as a matter of law. For multiple claims, AWS contends that JMOL is warranted based solely on Dr. Grama's testimony disputing opinions professed by Dr. Goodrich. See, e.g., Def.'s Rule 50(b) Mot. at 7 (hash function limitations); Id. at 9 (transfer limitations). "[B]ut the jury was in the best position to determine whether it found Dr. [Goodrich] or Dr. [Grama] more persuasive." Apple Inc. v. Samsung Elecs. Co., 839 F.3d 1034, 1045 (Fed. Cir. 2016). AWS can only prevail on its motion if it demonstrates that no reasonable jury could have decided as this jury did. AWS has failed to do so. This Court finds that the jury's findings on each of the limitations AWS identifies were supported by sufficient evidence.
a. Redirect and plurality of location servers
AWS contends that there was no evidence that DDB met the "client" and "redirect" limitations in claim 18 of the '640 patent and claim 10 of the '978 patent. Claim 18 requires that if a location server does not contain the requested location string, it "transmits a redirect message to the client" containing information that can be used to calculate the correct data location server. '640 pat., col. 22, lns. 62-65. Likewise, claim 10 requires that each location server is able to return a redirect message "compris[ing] information for finding a location server having location information related to the desired identifier." '978 pat., col. 26, lns. 36-45. AWS contends that Kove "failed to show that the redirect message contains information to calculate a location of a different data location server and information that contains a location string" and presented "no evidence showing the required transmission is 'to the client.'" Def.'s Rule 50(b) Reply at 7-8.
Dr. Goodrich testified that DDB infringes on claim 18 of the '640 patent and claim 10 of the '978 patent. For claim 18, he explained that "the client is the request routers. The location servers are the metadata nodes, and the data repository is the storage nodes." Tr. at 671:18-23. He stated that when a request router receives a request for an item in a table, the request router will first consider if the applicable location data is available in the local cache. If not, the request router will contact the metadata node, and the metadata node will send back to the request router metadata that includes the storage node ID (the location). He testified that the information that is sent back to the request router, the client, satisfies the "redirect message" requirement. Id. at 658:6659:3. Dr. Goodrich further testified that the information in the redirect message includes "a metadata node ID with a host and a port," which tells the client "which other metadata node now to go to finish this process." Id. He explained that this functionality was consistent with his analysis of DDB source code. Id. at 656:17-657:3. Dr. Goodrich relied on the same evidence when testifying about the "redirect message" limitation in claim 10. Id. at 679:22-680:9.
The only basis for overturning the jury's verdict that AWS offers is the testimony of Dr. Grama, who stated that, based on his analysis of the storage node source code, DDB does not utilize a "redirect message." See Id. at 1476:9-1479:11. But the jury was not required to credit Dr. Grama's testimony. Apple Inc., 839 F.3d at 1045. The Court finds that Kove presented sufficient evidence from which the jury could find that DDB satisfies the "client" and "redirect message" limitations.
AWS also relies on the purported lack of evidence of transmission of location information to support its argument that Kove failed to show that the accused products satisfy the "plurality of location servers" limitation. At the summary judgment stage, the Court construed the patents' "plurality of location servers" language to require location servers in a "non-hierarchical" configuration, meaning "any given server is able to return either the requested information or information useable by the client to locate the server with the requested information." See Jury Instructions at 17. Dr. Goodrich testified that within DDB, if a metadata node does not have the desired location information, it will send metadata to the request router that allows the client to find the correct metadata node. Tr. at 657:4-659:13. Likewise, for S3, Dr. Goodrich explained that any given BM is able to return requested location information, and if a KFC does not have the information necessary to resolve a get request, it will query a BM and then send the appropriate location information to the web server. Id. at 609:9-15; 609:19-610:4. Dr. Goodrich supported his testimony with an Amazon training video and S3 source code. See id. at 610:5-611:11.
AWS notes that Dr. Goodrich confirmed during cross-examination that KFCs do not store location information for every request. See Tr. at 840:10-14. But AWS does not explain how this testimony supports its argument regarding non-infringement. The non-hierarchical structure limitation is met if each location server is enabled to return the requested location information or information that the client can use to find the server that contains that information. Given Dr. Goodrich's testimony that a KFC maintains its ability to send location information to the web server even if the location information is not available in its local cache, see id. at 571:21-576:9; 1777:16-23, KFCs can meet the "non-hierarchical" configuration requirement even if they do not store the location information necessary to resolve every get request.
AWS does not acknowledge Kove's evidence and asks the Court to find in its favor solely on the basis of competing testimony from its own expert witness. But "[i]t was within the province of the jury to weigh the testimony presented by both sides and make its finding." GlaxoSmithKline LLC v. Teva Pharms. USA, Inc., 7 F.4th 1320, 1329 (Fed. Cir. 2021).
b. Non-hierarchical structure
In support of its argument that JMOL is warranted on various claim constructions related to the "non-hierarchical" structure required by the asserted patents, AWS rehashes its arguments regarding Dr. Goodrich's testimony related to the "plurality of location servers" claim limitation and his supposed admission that location servers do not store the necessary information to meet this claim limitation. The Court fully addressed these arguments above and need not repeat its legal analysis here. AWS also asserts, with no explanation or citation to supporting evidence, that Kove "failed to show" or "never showed" that the accused products satisfy various specific limitations. See Def.'s Rule 50(b) Mot. at 7-8. As the Seventh Circuit has stated, "it is not this court's responsibility to research and construct the parties' arguments." Draper v. Martin, 664 F.3d 1110, 1114 (7th Cir. 2011) (citation omitted); see also Illinois Tamale Co. v. El-Greg, Inc., No. 16 C 5387, 2019 WL 4395139, at *10 (N.D. Ill. Sept. 13, 2019) (Kennelly, J.) (finding Rule 50(b) movant's argument waived based on bare assertion that nonmovant had failed to meet its burden at trial). AWS's remaining arguments with respect to this limitation are underdeveloped and therefore waived. Trs. of Chicago Painters & Decorators Pension, Health & Welfare, & Deferred Sav. Plan Tr. Funds v. Royal Int'l Drywall & Decorating, Inc., 493 F.3d 782, 789 (7th Cir. 2007).
c. Hash function
Dr. Grama's testified that DDB "didn't apply a hash function to the partition key (the accused identifier) and thus didn't meet the 'based on a hash function' requirement of '170 patent claims 1 and 2." Def.'s Rule 50(b) Mot. at 7. AWS asserts that Kove failed to rebut Dr. Grama's testimony. But Kove's expert witness expressly rebutted Dr. Grama's opinion through his description of DDB functionality. Dr. Goodrich testified that DDB takes the partition key portion of a primary key and "put[s] it into a hash function." Tr. at 666:10-23. Dr. Goodrich then identified a portion of DDB source code that served as the basis for this opinion, stating that the partition key is the input for the hash function, and the output is the partition ID, which is used to identify the correct location server. Id. at 667:8-668:7. "[F]aced with competing expert testimony," the jury was free to credit Kove's expert over AWS's expert. Intell. Ventures I LLC v. Motorola Mobility LLC, 870 F.3d 1320, 1327 (Fed. Cir. 2017).
AWS also contends that S3 does not infringe on these claims because, as Dr. Grama testified, "KFCs didn't distribute iNodes according to a hash function." Def.'s Rule 50(b) Mot. at 7. The scope of claims 1 and 2 are not as narrow as AWS suggests. The relevant claim language reads, "the portion of the data location information included in a corresponding one of the data location servers is based on a hash function used to organize the data location information across the plurality of data location servers." Kove IO, Inc., 2021 WL 5988928, at *4. Judge Pallmeyer construed the "based on a hash function" language to include the requirement that the portion of the data location information is "based on a hash function that maps identifier strings to one or more of the data location servers." Id. at *15. AWS provides no support for its contention that a limitation requiring the use of a hash function to "organize" location information across location servers includes a requirement that the hash function be used to "distribute" the locations themselves. See id. ("Claim 1 is not limited to any particular manner of organizing data based on a hash function.").
Furthermore, Dr. Goodrich testified that S3 satisfies this claim requirement because within S3, a consistent hash function is used to map identifier strings to location servers and determine which KFC stores the desired location data. Tr. at 556:21-558:9; 599:10-24. Dr. Goodrich presented to the jury a comment on S3 source code files explaining that S3 "map[s] an item to a bucket by hashing the item." Id. at 598:10-24. The jury also viewed a presentation by Jason McHugh, an Amazon engineer, stating that hash functions are used to "determine where on the consistent hashing circle that machine lives." Id. at 597:6-11. Dr. Goodrich's testimony provided sufficient evidence for the jury to reasonably conclude that S3 meets the "based on a hash function" requirement.
d. Performance limit
AWS argues that Kove "failed to show" that the accused products infringe on claim 17 and 30 of the '978 patent. Def.'s Rule 50(b) Mot. at 9. The relevant portion of claim 17 requires each location server to be able to "transfer a portion" of identifiers and their associated locations to another location server "when a performance criterion of the first location server reaches a predetermined performance limit." '978 pat., col. 27, lns. 40-43. Likewise, claim 30 requires the automatic transfer of identifiers/locations when the first data location server reaches a predetermined performance limit. '978 pat., col. 28, lns. 37-44. AWS asserts that within DDB, data is transferred manually by engineers based on their technical judgment rather than according to a predetermined performance limit.
The trial record supports the jury's verdict. The jury viewed deposition testimony from AWS engineer James Sorenson, who testified that partitions within DDB can split into two. He stated that this process leaves "part of the data resid[ing] on the original node" while "part of the data will be moved to new nodes." Tr. at 683:15-684:3. The jury also saw deposition testimony from AWS engineer Donal Walsh, who stated that data can be transferred between metadata nodes for load balancing purposes. Id. at 684:8-22. Dr. Goodrich stated that load balancing "has to do with these predetermined thresholds that either happen because of transactions getting too hot or the disc drives getting too full." Id. at 684:23-685:1. He further explained that predetermined performance limits are demonstrated through the use of metadata alarms, which are triggered when input/output operations exceed a certain value or storage within the disc space reaches a certain capacity. Id. at 685:20-686:4. Dr. Goodrich also testified that based on his analysis of DDB source code, metadata is transferred among metadata nodes based on different performance levels that are met when a metadata node "fails, is non-responsive, is corrupt, is being retired or removed from the fleet, or is overloaded." Id. at 686:11-21. A reasonable jury could have concluded that the partition split process that occurs for load balancing purposes satisfies the "predetermined performance limits" requirement.
AWS also contends that Kove failed to present evidence of the same requirement being met in S3. AWS's sole piece of supporting evidence is one line from Dr. Grama's testimony where he stated, "this notion of predetermined performance limit triggering transferring does not happen either in KFCs." Id. at 1508:13-5; Def.'s Rule 50(b) Mot. at 9. Dr. Grama's non-infringement conclusions alone are insufficient for this Court to find in AWS's favor. See Milwaukee Elec. Tool Corp. v. Snap-On Inc., 288 F.Supp.3d 872, 882 (E.D. Wis. 2017) ("[Rule 50(b) movant] seems to think it can succeed if it shows merely that it could have convinced the jury to go its way, but this is no reason to disturb the jury's determinations.").
Based on the trial record, there is no basis for this Court to substitute its judgment for that of the jury. This Court therefore denies AWS's motion for judgment as a matter of law on non-infringement.
4. Dr. Goodrich's testimony
AWS argues that no reasonable jury could rely on Dr. Goodrich's opinions. First, AWS asserts that Dr. Goodrich "confirmed at trial" that he relied on other individuals to "review AWS's source code" and "draft the source code appendices to his [infringement] report." Def.'s Rule 50(b) Mot. at 9.
AWS provides an incomplete description of Dr. Goodrich's testimony. Dr. Goodrich testified that he asked Dr. Babb, a researcher at MIT, to review AWS's source code to "do a preliminary analysis that identified which source code files would be relevant." Tr. at 762:2-10. But after Dr. Babb completed his preliminary analysis, Dr. Goodrich spent two days reviewing the source code in person at AWS's counsels' office and then spent "additional time" reviewing the source code on a USB drive. Id. at 761:17-20. Furthermore, although Dr. Babb wrote the first draft of the source code appendix included in Dr. Goodrich's expert report, Dr. Goodrich stated that he edited that appendix "based on [his] analysis so that it comprises [his] opinions in this matter." Id. at 763:13-15; 764:7-11. AWS has not shown that this departed from typical practice for experts in similar patent infringement cases or, for that matter, other cases involving voluminous, complex, or technical evidence. See Zelinski, 335 F.3d at 640 (noting party's failure to present evidence that the expert's methodology was atypical). In any event, Dr. Goodrich fully explained his source code review process to the jury, and the jury ultimately found his testimony credible. It was not unreasonable for the jury to rely on Dr. Goodrich's source code analysis.
Second, AWS contends that Goodrich's testimony "consisted largely of conclusory statements" that various materials support his infringement contentions. Def.'s Rule 50(b) Mot. at 9. In support of this contention, AWS cites Goodrich's testimony confirming that he did not present to the jury any portion of DDB source code that refers to a primary key. See Tr. at 852:13-20. But this portion of Dr. Goodrich's testimony does not contain an assertion that the source code supports his infringement opinions, so it is unclear how it relates to AWS's argument regarding the "conclusory" nature of Dr. Goodrich's testimony. And Dr. Goodrich supported his primary key opinions using other sources of evidence, including internal Amazon documents. See id. at 638:10-639:5. Additionally, Dr. Goodrich reviewed AWS source code and explained how it supported his infringement conclusions at multiple points during his over ten hours of testimony. See, e.g., Id. at 667:8-668:7; 1172:24-1173:15.
Third, AWS argues that Kove's counsel improperly suggested that the jury should disregard the source code. AWS's description of events is incomplete. During closing argument Kove's counsel made the following statement:
Their case in many ways comes down to this whole say-so thing, telling you what to think without showing you the evidence. And even the source code. The source code is the best evidence. I agree. Sure is. But guess what? Y'all can't read it. I can't read it either, just for the record. Okay? But they can and they went through it, and there's nothing in there that's inconsistent with the documents. If these documents were wrong in some material way, wouldn't they have told you that? Wouldn't they have pointed it out? Wouldn't they have used some of these documents?Id. at 2016:7-17. Kove's counsel did not suggest that the jury should ignore the source code when weighing the evidence in the case. Rather, counsel simply acknowledged that many, if not all, of the jury members lacked the technical expertise necessary to fully comprehend the source code and argued that the testimony of AWS engineers who did review the source code was consistent with various documents presented to the jury. AWS does not explain how it was prejudiced by this argument, nor does it present any evidence from which this Court could find that "there is a reasonable probability that the verdict of a jury has been influenced" in an inappropriate way. DeRance, Inc. v. PaineWebber Inc., 872 F.2d 1312, 1327 (7th Cir. 1989) (quoting City of Cleveland v. Peter Kiewit Sons' Co., 624 F.2d 749, 756 (6th Cir. 1980)). The jury was instructed that the attorneys' opening and closing arguments are not evidence. Jury Instructions at 3. Furthermore, multiple witnesses acknowledged that the source code is important evidence, and one of Kove's expert witnesses referred to the source code as the "ground truth." Tr. at 774:22-775:3. And AWS did not object to counsel's statements during the trial. "Having failed to seek a cure at the time, it is in a poor position to object to an allegedly improper remark in its post-trial briefs." Third Wave Techs., Inc. v. Stratagene Corp., 405 F.Supp.2d 991, 1010 (W.D. Wis. 2005).
Finally, AWS contends that Dr. Goodrich overstated the extent to which his descriptions of the accused products were consistent with the testimony of AWS engineers Dr. Allan Vermeulen, Seth Markle, and Timothy Rath. For example, AWS points to Dr. Goodrich's testimony that it was "undisputed" that each iNode contains a Blindex. See Tr. at 1745:5-8. But testimony from AWS engineers support rather than undermine Dr. Goodrich's testimony. Dr. Vermeulen testified that "the iNode is actually the data structure that contains the Blindex." Id. at 1290:17-1291:13. Markle also confirmed that the Blindex is the "most important field" within an iNode. Id. at 1364:711. AWS also directs the Court's attention to Dr. Goodrich's statement that he agreed with AWS witness testimony that KFCs provide the correct information if they are alive and respond to a request. See id. at 1753:9-16. But AWS does not cite any statements from Dr. Goodrich that contradict his testimony. And Dr. Goodrich's testimony that KFCs respond to requests by either providing the volume ID and blob ID of the data or with the information they receive from a BM isn't inconsistent with the notion that a functioning KFC will always provide the correct answer to a data request. See id. at 561:2-562:4.
Kove's counsel attributed this statement to Markle, but Rath is the AWS engineer that testified that "as long as the KFC is alive and responds," it provides the "correct answer" to a request for data. Id. at 1384:17-1385:3.
AWS also takes issue with Goodrich's statement that hearing testimony from AWS engineers about how the accused products operate did not change his infringement opinions. See id. at 1741:6-11. But Dr. Goodrich explained that his infringement opinions remained the same because the engineers did not analyze the Court's claim constructions and "admitted that they didn't read the patents." Id. at 1741:12-17. Furthermore, as noted above, Dr. Goodrich's disagreement with AWS's engineers does not show that his testimony is inherently unreliable or that a jury would be unreasonable to credit his opinions. The jury reasonably could have decided that, because the AWS engineers did not read the patents, the persuasive value of their testimony on infringement issues was limited. Alternatively, the jury reasonably could have come to the independent conclusion that, despite AWS's witnesses' assertions to the contrary, their testimony did provide support for Goodrich's infringement opinions. For example, Dr. Vermeulen testified that the Blindex is a "label that describes the data." Tr. at 1292:22-1. He also testified that the Blindex contains both a volume ID and a blob ID, Id. at 1314:17-25, and that the volume ID "is the data in the Blindex that allows you to find the correct R2D2s" where the data is stored. Id. at 1350:3-12. Based on Dr. Vermeulen's testimony that the Blindex contains information about the R2D2 where the data is stored, the jury reasonably could have (and based on the verdict, apparently did) conclude that the Blindex was more than simply a "label."
"When reviewing a jury verdict, the question is not whether the jury believed the right people, but only whether it was presented with a legally sufficient amount of evidence from which it could reasonably derive its verdict." Zelinski, 335 F.3d at 638 (citation omitted). The jury reasonably relied on Dr. Goodrich's testimony in reaching its verdict.
5. Damages
AWS argues that Kove did not provide sufficient evidence to support the jury's damages award. The Federal Circuit has stated that when "reviewing damages awards in patent cases," a court must "give broad deference to the conclusions reached by the finder of fact." Monsanto Co. v. McFarling, 488 F.3d 973, 981 (Fed. Cir. 2007). The Federal Circuit has also instructed that a jury's damages award "must be upheld unless the amount is grossly excessive or monstrous, clearly not supported by the evidence or based only on speculation or guesswork." Id. (citation omitted). AWS contends that judgment as a matter of law should be awarded because Bergman failed to properly apportion his damages calculation to exclude the value of non-patented features in the accused products, improperly relied on an AWS project document for his damages analysis, and did not sufficiently explain the basis for his bargaining split opinions.
a. Apportionment
Under 35 U.S.C. § 284, "damages awarded for patent infringement must reflect the value attributable to the infringing features of the product, and no more." Commonwealth Sci. & Indus. Rsch. Organisation v. Cisco Sys., Inc., 809 F.3d 1295, 1301 (Fed. Cir. 2015) (internal quotation omitted). This aspect of damages law is reflected through the principle of apportionment, which requires "separat[ing] the value of the allegedly infringing features from the value of all other features." Id.
AWS contends that Bergman's damages analysis contained incorrect assumptions about S3's functionalities. But each of the allegedly inaccurate assumptions AWS raises are explained and supported by Bergman's reliance on Kove's experts to gain a better understanding of the features of the accused products.
To analyze the technical benefits that the Kove patents provided to the accused products, Kove's experts considered the "non-infringing alternative" that AWS would need to implement to avoid infringing on the patents. Tr. at 701:11-24. AWS asserts that Bergman's non-infringing alternative improperly required customers to manage their own data because he "incorrectly assumed that S3's ability to scale derives solely from infringing features." Def.'s Rule 50(b) Mot. at 12. Bergman testified that his understanding of the scaling ability of KFCs and BMs came from Dr. Goodrich. Tr. at 1186:17-24. Bergman testified that this understanding was that "without access to the KFC scaling claims[,] . . . AWS would not have access to the KFC cache and . . . all of the [data] requests would go to the brick manager." Id. at 1047:25-1048:7. This understanding is the basis for Bergman's conclusion that without scaling claims, S3 would no longer be able to offer automatic partitioning and customers would have to partition their data manually. Id. at 1069:3-21; 1076:2-10. Bergman further testified that removing the benefits of the scaling claims from S3 would result in "increased costs, decreased availability, increased latency, and a decrease in system durability." Id. at 1048:23-1049:2. Bergman also stated that his understanding is that KFCs are only able to locate iNodes within the cache due to benefits derived from the patent hashing claims. Id. at 1047:17-24. He explained that without the use of hash functions or hash tables to distribute iNodes, no requests would reach the KFC cache and all requests would therefore go to the BMs. Id. at 1047:25-1048:7. Bergman's understanding of S3's functionality is consistent with Dr. Goodrich's testimony. Id. at 702:14-703:4 (explaining benefits of hashing claims for S3); 713:16-24 (describing consequences of removing KFC cache); 731:3-732:13 (explaining impact of scaling claims on S3 functionality); 732:15-733:22 (describing consequences of removing scaling claims).
Similarly, Dr. Goodrich is the source of Bergman's underlying assumptions regarding "how often KFCs in S3 transfer iNodes when attributing damages to the KFC cache." Def.'s Rule 50(b) Mot. at 12. Dr. Goodrich testified that Amazon's internal testing revealed that 80% of requests that reach the KFC level are resolved by the KFC cache, while the remaining 20% go to the BM level. Id. at 707:11-21. Bergman explained that to develop a non-infringing alternative, he worked with Dr. Goodrich to stimulate a model where the "cache hit ratio," or the percentage of requests that are resolved by the KFC cache without reaching the BMs, decreased from 80% to 0%. Id. at 1053:6-1-54:6. Dr. Grama also used this 80% hit ratio number when explaining the functionality of the KFC cache. Id. at 1503:3-12. Although AWS may have offered contradicting evidence about S3's functionality, Bergman was entitled to rely on Dr. Goodrich's expert opinions to develop his damages analysis.
AWS argues that Bergman failed to properly apportion damages to DDB because it was based on the incorrect assumption that 100%, rather than .25%, of get requests infringe. Kove argues that AWS forfeited this argument by failing to include it in its Rule 50(a) motion. This Court agrees. The Seventh Circuit has stated that "[b]ecause the Rule 50(b) motion is only a renewal of the preverdict motion, it can be granted only on grounds advanced in the preverdict motion." Thompson v. Mem'l Hosp. of Carbondale, 625 F.3d 394, 407 (7th Cir. 2010) (citation omitted); Andy Mohr Truck Ctr., Inc. v. Volvo Trucks N. Am., 869 F.3d 598, 605 (7th Cir. 2017) ("[I]f a party raises a new argument in its Rule 50(b) motion that was not presented in the Rule 50(a) motion, the non-moving party can properly object."). At trial, AWS moved for judgment as a matter of law under Rule 50(a). In its Rule 50(a) motion, AWS argued that Bergman "failed to explain how he apportioned damages to the allegedly infringing features." Def.'s Rule 50(a) Mot. at 14. But the only examples of this failure that AWS cited were related to S3, not DDB. Specifically, AWS asserted that Bergman had incorrectly attributed the total value of S3's KFC cache to the '170 patent claims and omitted S3's index system from his non- infringing alternative for other claims. See Def.'s Rule 50(a) Mot. at 14. AWS did not discuss get requests or DDB in its Rule 50(a) motion and this argument is therefore forfeited. Webster v. CDI Indiana, LLC, 917 F.3d 574, 578 (7th Cir. 2019).
The Court finds AWS is not entitled to judgment as a matter of law with respect to damages based on Bergman's apportionment opinions.
b. Goldilocks
To determine the value of the scaling claims to S3, Bergman reviewed Amazon documents related to a storage class product called "Goldilocks." S3 contains multiple storage classes, including the standard infrequent access storage class (SIA) and the one zone IA storage class (one zone). Tr. at 1615:4-25. While the SIA stores data in multiple zones, the one zone stores data within a single zone. Id. Goldilocks is an SIA product that AWS developed as a potential competitor to a Google product called Nearline. Id. at 1070:10-1071:2; 1681:1-18. The jury viewed an Amazon document indicating that AWS estimated that Goldilocks merited a 20 to 30 percent price premium over Nearline. Id. at 1071:3-15. The same document also stated that the price premium was "warranted" due to the "AWS brand" as well as "slight product differences." Id. Bergman concluded that Goldilocks's seamless tiering functions, which eliminated a customer's need to manually partition their own data, accounted for a 20% price difference between Goldilocks and Nearline. Id. at 1070:10-1071:2.
To gain a better understanding of Goldilocks's features and how they relate to the patents' scaling claims, Bergman relied on Kove experts Karim Fanous and Dr. Goodrich. Dr. Goodrich testified that if S3 were unable to scale within the BMs, the alternative would be a process called "manual partitioning," which would require customers to manually track, transfer, and manage their own data rather than having S3 automatically partition the data. Id. at 733:9-22. Dr. Goodrich further testified that he considered manual partitioning an effective means of approximating the benefits of the patents' scaling claims. Id. at 733:23-734:1. Dr. Goodrich explained that he compared a Goldilocks to a manual portioning service in order to assess the effects of removing the scaling claims from S3. Id. at 733:23-735:15. Similarly, Fanous testified that although there is no non-scalable version of S3, he approximated the pricing differences between a manual and automatic partitioning system by comparing Google Nearline to Goldilocks. Id. at 454:3-20.
AWS asserts that Bergman "failed to consider" various differences between Goldilocks and Nearline in his analysis. Def.'s Rule 50(b) Mot. at 13. But Bergman testified that it was the "compilation of all of the evidence," rather than solely AWS's estimate, that informed his conclusions about the price difference between Goldilocks and Nearline. Tr. at 1200:13-1201:2. While testifying, Bergman reviewed multiple Amazon documents that described features of S3 and Goldilocks. He stated that one of the "key features" of Goldilocks was "seamless tiering," which allows the system to "move data within a storage class without having the customer have to do it." Id. at 1070:2-16. Bergman also cited an Amazon document identifying seamless tiering as "one of our key differentiators from Nearline." Id. at 1070:10-1071:2. The same document also highlighted seamless tiering as a "core feature" of Goldilocks. Id. Bergman cited an additional Amazon document stating, "[w]e believe our proposed product definition is slightly better than Google's Nearline product because customers do not have to change their application to point to a new bucket." Id. at 1071:3-15.
Fanous's testimony also supported Bergman's conclusions. Fanous testified that he considered multiple differences between Goldilocks and Nearline and concluded that "storage class" was the biggest distinction between the two products. Id. at 455:4-7. He explained that although both products moved objects across tiers, in Goldilocks, "the movement happens automatically," and in Nearline, "the end user has to move her data" manually in a manner "similar to some of the operations you would be doing if you were manually partitioning." Id. at 455:8-19. Furthermore, S3 manager Christoph Bartenstein testified that scalability is an important feature for S3 customers. Id. at 1633:3-9. Testimony from both parties' witnesses supported Bergman's conclusion that seamless tiering was a core feature of Goldilocks and contributed to its price premium over Nearline.
To the extent AWS argues that Bergman should have considered the impact of other features on the price difference, AWS is simply rehashing a contention that was presented to the jury at trial. During cross-examination, Bergman was asked about an AWS document that stated listed multiple features that AWS believed supported Goldilocks's pricing premium, including "S3 lifecycle, IAM, KMS, and other AWS integration with S3." Id. at 1197:2-7. Bergman stated that he relied on other evidence that describes seamless tiering as the core difference between Nearline and Goldilocks. AWS's witnesses also questioned Bergman's methodology. Bennis testified that Bergman's damages model was "inherently overstated" because he failed to account for "other elements that help drive the price premium." Id. at 1681:1-24. The jury was entitled to credit Bergman's testimony over Bennis's in reaching its verdict. Rehco LLC v. Spin Master Ltd., No. 13-CV-2245, 2020 WL 7025091, at *7 (N.D. Ill. Nov. 30, 2020) ("[T]he jury was free to accept or reject aspects of the evidence and to credit the testimony it found most credible.").
c. Bargaining split
The most common approach to determining reasonable royalty damages is the hypothetical negotiation approach. Lucent Techs., Inc. v. Gateway, Inc., 580 F.3d 1301, 1324 (Fed. Cir. 2009). This approach "attempts to ascertain the royalty upon which the parties would have agreed had they successfully negotiated an agreement just before infringement began." Id. at 1324. Developing a hypothetical negotiation analysis "necessarily involves an element of approximation and uncertainty." Id. at 25 (citation omitted). That said, the Federal Circuit has cautioned that damages "must not be left to conjecture by the jury. They must be proved, and not guessed at." Promega Corp. v. Life Techs. Corp., 875 F.3d 651, 660 (Fed. Cir. 2017) (quoting Philp v. Nock, 84 U.S. 460, 462 (1873)).
One method of calculating a reasonable royalty rate is "through a proper analysis of the Georgia-Pacific factors." Exmark Mfg. Co. v. Briggs & Stratton Power Prod. Grp., LLC, 879 F.3d 1332, 1349 (Fed. Cir. 2018) (citing Georgia-Pacific Corp. v. U.S. Plywood Corp., 318 F.Supp. 1116 (S.D.N.Y. 1970)). "The standard Georgia-Pacific reasonable royalty analysis takes account of the importance of the inventive contribution in determining the royalty rate that would have emerged from the hypothetical negotiation." Id. (citation omitted).
Bergman testified that his analysis covered a damages period from 2012 until the patents' expiration in 2018. When asked which Georgia-Pacific factors he considered, he carefully explained the four factors that were related to his bargaining split analysis: quantified benefits, unquantifiable benefits, benefits to other AWS products, and AWS's typical licensing practices. Tr. at 1095:24-1096:19. Bergman stated that the quantified benefits included the $1 billion in profit that he contended Amazon earned from Kove's patents during the damages period. Id. at 1097:6-8. For unquantified benefits, Bergman referenced Dr. Goodrich's testimony regarding the benefits that the Kove patents provided to AWS's products. See id. at 689:22-700:23 (explaining technical benefits that the patents-in-suit provide to AWS). Bergman noted that "the parties of the hypothetical negotiation would know about these benefits and would take them into account." Id. at 1097:13-1098:5.
Regarding benefits to other AWS services, Bergman explained that he considered the extent to which AWS's use of S3 is "driving business in other [AWS] services." Id. at 1098:14-1099:5. Bergman quoted from multiple AWS documents discussing how S3 drives customers to use other AWS services. For example, Bergman read from a 2012 AWS document describing how customers tend to "move their compute," or their computing resources, "close to their data, which will generate additional revenue and margin" for AWS. Id. at 1099:6-20. He also read from another AWS document that stated, "for each terabyte of data in S3 we sell, we generate direct revenue of $22 based on public pricing and up to $31 in indirect revenue." Id. at 1100:24-1101:9. An additional AWS document Bergman reviewed stated that ""[c]ustomers who spend $1 storing data in AWS spend 4 to $7 on other AWS services." Id. at 1101:10-1101:21. When asked how the additional four to seven dollars AWS received every $1 spent on S3 storage impacted his quantified benefits analysis, Bergman stated that AWS received "between 4 billion and 7 billion dollars in additional profit from other services" as well as the "billion dollars in additional benefit directly attributable to S3 and DDB." Id. at 1101:22-1102:6
Bergman further testified that he evaluated AWS's investment decision-making process by viewing deposition testimony from AWS employees indicating that the company prioritizes potential growth over direct profits. Id. at 1102:7-1104:4. He also showed the jury a quote from Swami Sivasubramanian, DDB's general manager, who stated that AWS is willing to develop technology "even if it results in zero" as long as it addresses a "critical need" for AWS customers. Id. at 1103:17-22.
Bergman concluded that:
[T]aking into account the fact that this is profit that would basically disappear but for a license, understanding the significant downstream impacts associated with S3 revenue, those unquantifiable benefits that we've been talking about this entire time, as well as an understanding of how AWS typically goes about making investment decisions, it's my opinion that it would be at least 50 percent and even all the way up to a hundred percent of that quantifiable benefit.Id. at 1104:5-15.
AWS argues that Bergman failed to sufficiently support his bargaining split opinions. Specifically, AWS challenges Bergman's testimony that in a hypothetical negotiation "the parties would've agreed that Kove would receive at least 50%, and up to 100%, of the incremental profits attributable to the use of the patents-in-suit." Def.'s Mot. for New Trial at 15. In support of its argument, AWS cites the Federal Circuit's decision in Exmark. In that case, the defendant argued that the plaintiff's damages expert's opinion should have been excluded because she did not sufficiently explain how she reached her proposed bargaining split. Exmark Mfg. Co., 879 F.3d at 1347. At trial, the expert discussed each of the Georgia-Pacific factors, explained her understanding of the benefits that the asserted patent provided to the accused product, and stated that negotiators would have recognized these benefits. Id. at 1349-50. She also noted that these benefits were related to Georgia-Pacific factors nine and ten. Id. at 1350. The expert subsequently "concluded with little explanation" that the parties would have agreed to a 5% reasonable royalty rate. Id. at 1349.
The Federal Circuit held that "a superficial recitation of the Georgia-Pacific factors, followed by conclusory remarks, [cannot] support the jury's verdict." Id. at 1350 (citation omitted). The court also noted that "[w]hile mathematical precision is not required, some explanation of both why and generally to what extent the particular factor[s] impact[ ] the royalty calculation is needed." Id. at (citation omitted). The Federal Circuit concluded that the district court erred by admitting the damages expert's report because it "failed to adequately tie the expert's proposed reasonable royalty rate to the facts of this case," and the court ordered a new trial on the issue of damages. Exmark Mfg. Co., 879 F.3d at 1349.
In this case, by contrast, the Court overruled AWS's pretrial motion seeking to exclude Bergman's bargaining split testimony, concluding that his testimony on that point was sufficiently explained in his written report. His testimony at trial, though not as detailed as the extended discussion of the relevant considerations in his Rule 26(a)(2) report-likely, in part, due to time limits imposed on the parties' trial presentations-laid out in detail the relevant factors and explained how they factored into his ultimate conclusion regarding the bargaining split. AWS focuses on Bergman's opinion that in a hypothetical negotiation, Kove would end up with a royalty consisting of at least 50% and as much as 100% of the incremental profits from AWS's use of the patents, arguing in substance that this cannot possibly make sense. But that is a bit of a diversion, because the baseline number does not account for the significant, and much larger, benefit that AWS derived from its use of the patents that is not captured in the profit figure. As Kove notes, this included, according to Bergman, $4 billion to $7 billion In profits from convoyed and add-on sales attributable to AWS's infringement. See Tr. at 1101:10-1102:6. Bergman then relied on testimony from AWS decision makers to the effect that AWS would be willing to give up its quantified profits from direct infringement for the long-term gain from its use of the Kove patents. In other words, Bergman explained how he reached his bargaining split opinion; he did not simply pull it out of the air. It is true that he did not reduce it to a formula. But neither Exmark Manufacturing Co. or any other case requires that sort of precision; we are talking about a hypothetical negotiation here, not a mathematical formula or a chemical equation. The Federal Circuit has made it clear that assessment of damages "does not demand absolute precision but may involve some degree of approximation and uncertainty." EcoFactor, Inc. v. Google LLC, 104 F.4th 243, 254 n.6 (Fed. Cir. 2024) (internal quotation marks omitted).
Given that Bergman fully examined and explained the Georgia-Pacific factors as applied to the circumstances of the case, this is not a situation in which a party's damages expert "plucked" its proposed royalty rate "out of nowhere." Exmark Mfg. Co., 879 F.3d at 1351.
For the foregoing reasons, the Court denies AWS's motion for judgment as a matter of law with respect to damages.
B. AWS's motion for a new trial
AWS alternatively moves for a new trial on all of Kove's claims and, alternatively, on the question of damages. "A motion for a new trial can be granted when the district court-in its own assessment of the evidence presented-believes that the verdict went against [its] manifest weight." Abellan v. Lavelo Prop. Mgmt., LLC, 948 F.3d 820, 831 (7th Cir. 2020) (citation omitted); see also Estate of Burford v. Accounting Practice Sales, Inc., 851 F.3d 641, 646 (7th Cir. 2017) ("A new trial should be granted only when the record shows that the jury's verdict resulted in a miscarriage of justice or where the verdict, on the record, cries out to be overturned or shocks our conscience."). "[A] court will set aside a verdict as contrary to the manifest weight of the evidence only if no rational jury could have rendered the verdict." Whitehead v. Bond, 680 F.3d 919, 928 (7th Cir. 2012).
1. Claim construction
a. Location
As noted above, the parties agreed prior to claim construction that a "location" within the meaning of the Kove patents is an encoding that "specifies where data pertaining to the entity identified by the identifier is stored." Kove IO, Inc., 2021 WL 5988928, at *7. In support of its motion for summary judgment, AWS argued that an iNode could not satisfy the parties' agreed upon construction of "location" in part because "the Blindex," a component within iNodes, "always stays the same once it has been created." Def.'s Mot. for Summ. J. at 25. Kove argued in its response brief that AWS's argument improperly narrowed the scope of the claim terms. See Pl.'s Resp to Def.'s Mot. for Summ. J. at 9-10. Because this Court partially adopted AWS's claim construction proposal, it did not address the parties' dispute over the "location" requirement at the summary judgment stage. Kove IO, Inc., 2024 WL 450028, at *11.
In its Motion in Limine No. 7, Kove sought to exclude evidence related to AWS's location-based non-infringement arguments at trial. Specifically, Kove objected to AWS's contentions that (1) a location must be "all that you need to find exactly the data that you're looking for" and (2) "something cannot be a location if the data moves but the location stays the same." Dkt. no. 808 at 16. After the final pretrial conference, AWS abandoned the first contention and stated that "Dr. Grama won't offer any opinion that a location must be all that you need to find exactly the data you're looking for." Dkt. no. 837 at 2. On AWS's second contention, the Court argued that AWS's position was contrary to the construction adopted by Judge Pallmeyer and that AWS had forfeited any argument for its preferred construction when it agreed to the construction that Judge Pallmeyer adopted. Id. at 3. This Court concluded, however, that AWS could advance the argument that "a Blindex is not location data because it does not help identify (or specify) where data is stored." Id.
AWS asserts that it was unfairly prejudiced by the exclusion of Dr. Grama's testimony in relation to whether a Blindex can provide location information given that it is a static component that does not change when the data moves. In support of its argument, AWS cites Innogenetics, N.V. v. Abbott Laboratories, 512 F.3d 1363 (Fed. Cir. 2008). In that case, the Federal Circuit reviewed a district court's determination that an expert's testimony rested on an incorrect understanding of the court's construction of the word "genotyping." Id. at 1376. The court had construed "a method of genotyping" as a method that "distinguishes" among types of hepatitis C virus (HCV) and "classifies" the HCV. Id. at 1377. During his testimony, the expert stated that a patent disclosed methods that "distinguish" between groups of HCV. Id. at 1376. Opposing counsel objected to this testimony, arguing that the expert's use of the term "distinguish" was improper because the expert, in his report, defined a "method of genotyping" as "the process of detecting and classifying" different HCV strains. Id. The district court found that the expert's testimony rested on an incorrect understanding of the construction of the word "genotyping" and granted JMOL on that basis. Id. The Federal Circuit reversed, noting that the dispute "seem[ed] to turn upon [the expert's] use of the words 'detect' and 'classify' and the district court's use of the words 'distinguish' and 'classify' in its construction of a 'method of genotyping.'" Id. at 1377. Because it "discern[ed] no difference in meaning between the definition of genotyping used by [the expert] and the one adopted by the district court," the Federal Circuit found that the expert witness's report and trial testimony were both consistent with the district court's claim construction. Id. at 1377-78.
This case is distinguishable. The Court's ruling on the exclusion of Dr. Grama's testimony was based on the substance of his opinions, not the particular language he used. In particular, the Court found that AWS's assertion that an encoding cannot specify where data is stored if it remains static when the data moves "is contrary to- because it imposes an additional limitation upon-the construction adopted by Judge Pallmeyer." Dkt. no. 837 at 2-3. Furthermore, AWS does not point to any evidence in the record to support a finding that the construction for which it advocates is required by the language of the claims. In support of its argument, AWS only cites the abstract of the '978 patent, which states that "[t]he data location is a key question when a distributed database has highly dynamic, and even spontaneous, data distribution properties." See Def.'s Mot. for New Trial at 3-4 (citing'978 pat., col. 1, lns. 65-67). This general description of the importance of locating data in a distributed database system in the "background" section of the patent is insufficient to import a new limitation into the claims.
Neither the claims nor the specifications of the patents-in-suit indicate that Kove intended to limit a "location" solely to components that change when the data moves. And, as AWS notes, an expert's infringement opinion is only admissible to the extent that it conforms with the Court's construction of the claims. See Phillips v. AWH Corp., 415 F.3d 1303, 1318 (Fed. Cir. 2005) ("[A] court should discount any expert testimony that is clearly at odds with the claim construction.") (internal quotation omitted); see also Def.'s Mot. for New Trial at 2 (citing Innogenetics, N.V., 512 F.3d at 1377). In his expert report, Dr. Grama asserted that a Blindex "does not change even when the location of the associated data on S3 changes, and so is not location data." Dkt. no. 808, Ex. 7 ¶ 265. This Court excluded Dr. Grama's opinion because it added an additional limitation to the parties' agreed upon construction of the term "location," not due to a mere dispute over "semantics." See Innogenetics, N.V., 512 F.3d at 1377. Furthermore, the jury heard testimony confirming that a Blindex is static. Dr. Goodrich acknowledged that a Blindex stays the same when the data moves. See Tr. at 1258:14-21 ("And the R2D2s that are associated with that volume ID and the blob ID can change, but the volume ID and the blob ID stay the same."). In sum, AWS was not prejudiced, let alone unfairly prejudiced, by the exclusion of testimony that contradicted the Court's construction of the asserted claims.
AWS contends that Dr. Goodrich's trial testimony provides further support for its argument that a Blindex cannot be a location. In response to a question about why he excluded SkyNet from his analysis, Dr. Goodrich responded:
[W]e actually have to go back to the definition of location .... So just to kind of like set the stage a little bit, there's lots of different locations that we can be talking about and what location means.
So, for example, in the physical world, the location where I'm staying, my hotel room is in the Palmer House hotel. I'm not going to tell you my room number, but it's also a room number inside the Palmer House hotel. So that character string, Palmer House, those letters, and my room number is a location. That's where I'm staying. That is a location. But it's what we would refer to as, I think, a logical location. Just that string by itself doesn't tell you how to get to my hotel room.
So point of that is that from the Court's claim construction that we have here in the juror notebook, what we're calling a location doesn't have to have all the information you need in order to find the requested data. It just has to be able to specify where data pertaining to the entity is stored. So you can use other things to get you there, but that information has to be able to help you find it. Otherwise, it's not a location.Tr. at 1255:6-12:1258:21.
AWS states that "S3's Blindex fundamentally differs from the hotel room number in Dr. Goodrich's example." Def.'s Mot. for New Trial at 6. AWS misunderstands Dr. Goodrich's analogy. Dr. Goodrich used this analogy to illustrate the difference between a layperson's definition of location and the "logical location" that is disclosed by the patents. For the purposes of his analogy, he testified that "the character string, Palmer House, those letters, and my room number is a location." Tr. at 1255:15-20. In other words, Dr. Goodrich suggested that the combination of the name of the hotel, Palmer House, and his room number at the hotel served as the "location" in order to illustrate that, as AWS conceded prior to the pre-trial conference, an entity can constitute a "location" under the Court's claim construction even if it does not provide all of the information needed to locate the data.
AWS asserts that Kove's definition of "location" is too broad. But AWS does not acknowledge that the definition of "location" it advocates for is quite narrow. Building off of Dr. Goodrich's hotel analogy, if Dr. Goodrich was assigned to room 801 on the eighth floor of the Palmer House, but then is moved to the room next door, room 802, then his room number would change, but his floor number would stay the same. If the Court adopts AWS's argument regarding the scope of the term "location," then the floor number of Dr. Goodrich's room cannot be a location because it stayed the same even while his room number changed. Although this result may not be consistent with the ordinary meaning of "location," AWS has not demonstrated that it is inconsistent with the language of the asserted claims. And whether the floor number (or, in technical terms, the iNode containing a Blindex) "specifies where the data is stored" is a question of infringement, not claim construction.
AWS also challenges the Court's sustaining of Kove's objection to Dr. Grama's testimony regarding whether a storage node ID constitutes a "location." Dr. Grama testified that a storage node ID does not satisfy the claims' "location" requirement because it provides only a partition, and "if I tell you where an entire partition is, you have absolutely no way of locating where a single row in that table is." Tr. at 1462:2-9. Dr. Grama's suggestion that the identification of a partition alone is insufficient to locate the data runs directly afoul of AWS's commitment not to offer testimony that a location must be "all that you need" to find the desired data. AWS confirms this by arguing in its reply brief that "under the controlling 'location' construction, the 'location' must at least specify that row of data." Def.'s Mot. for New Trial Reply at 2. But AWS agreed that it would not offer an argument at trial that a location must be all you need to locate the data. AWS "offers no reason for relieving it of the consequences of its own agreement." Third Wave Techs., Inc., 405 F.Supp.2d at 999. Dr. Grama's testimony was properly excluded.
The same motion in limine ruling provides the basis for precluding AWS's questioning regarding a statement that Dr. Goodrich made to the USPTO during the reexamination process. At trial, AWS asked Dr. Goodrich whether it was his opinion that the location is "updated if data is moved in the data repositories." Tr. at 808:2-4. AWS asked this question with the intention of arguing that although Dr. Goodrich testified that a Blindex can provide location data despite never changing, he "provided the opposite opinion at the Patent Office on what the patent means" when it refers to a location. Id. at 816:17-22. AWS argues that the Court should have allowed it to present Dr. Goodrich's allegedly inconsistent prior statements. But AWS acknowledged at trial that whether or not the location changes when data moves "was the subject the motion in limine was about." Id. at 817:6-8. As the Court explained at trial, allowing this line of questioning would have allowed AWS to "get in through the back door" a point that it "couldn't bring in through the front door." Id. at 817:9-15. In other words, AWS could not circumvent the Court's motion in limine ruling excluding its contention that a location must change when the data moves by cross-examining a witness about that same contention.
AWS asserts that Dr. Goodrich and Kove improperly "implied to the jury" that its non-infringement theory "relied on the Blindex merely not including 'extra information.'" Mot. for New Trial at 6. In both portions of testimony that AWS cites, Dr. Goodrich explain that he did not find it necessary to include SkyNet in his infringement analysis because he found that the Blindex was sufficient to meet the "location" limitation of the asserted claims. See Tr. at 1257:15-1258:3; 810:21-25. This testimony does not support AWS's assertion that Dr. Goodrich misrepresented its non-infringement contentions to the jury. And Dr. Goodrich made it clear that he considered whether the alleged location specifies where data is stored when developing his analysis. See Id. at 585:22-586:4 ("The Blindex tells us where the data is stored for a given object within the storage layer."). AWS points to no testimony from Dr. Goodrich suggesting that he applied a definition of "location" that is inconsistent with the Court's claim construction.
AWS also contends that the Court modified the "location" construction at the end of the trial by modifying the jury instructions. "An erroneous instruction regarding claim interpretation that affects the jury's decision on infringement is grounds for a new trial." Ecolab, Inc. v. Paraclipse, Inc., 285 F.3d 1362, 1373 (Fed. Cir. 2002). "[T]o warrant a new trial, [AWS] must show that the erroneous jury instruction was in fact prejudicial." Id. at 1374. If the error "could not have changed the result, the erroneous instruction is harmless." Id. (quoting Environ Prod., Inc. v. Furon Co., 215 F.3d 1261, 1266 (Fed. Cir. 2000)).
During the charge conference, the Court accepted Kove's proposal to add to the claim construction provided in the jury instructions that "[t]here is no requirement that the 'location' have all the information necessary to find and retrieve the requested data." Tr. at 1869:20-1870; see Jury Instructions at 15. This language was added to reaffirm, not modify, the Court's existing claim construction. AWS agreed before trial that it would not argue that a "location" as defined by the patents must be "all that you need to find exactly the data that you're looking for." Dkt. no. 837 at 2. AWS's argument that the language in the jury instructions indicating that the location does not need to have all of the information necessary to locate the data is somehow materially different from the Court's statement that a location need not have all the information needed to locate the data lacks merit. And as noted above, AWS has not shown that the claim construction on which the instruction was based was erroneous.
As the Court explained in denying motion for judgment as a matter of law, ample evidence was presented at trial to support a finding that the iNode is a location. Dr. Goodrich acknowledged that the Blindex does not change even if the data moves. Id. at 1258:14-21 ("[T]he R2D2s that are associated with that volume ID and the blob ID can change, but the volume ID and the blob ID stay the same."); 807:2-23. He opined, however, that the volume ID and the blob ID within the Blindex qualify as a "location" within the meaning of the patent claims. Id. at 805:1-5. At trial, Dr. Grama argued that the iNode cannot be a location because the Blindex does not contain location data. See, e.g., Id. at 1493:18-1494:1 (testifying that the Blindex source code does not indicate that it helps specify where data is stored). The jury's resolution of this conflicting testimony in Kove's favor was not against the manifest weight of the evidence.
b. Non-hierarchical
AWS contends that the jury's infringement finding was based on an erroneous construction of "non-hierarchical" and that the Court's construction allowed Kove to recapture claim language that it had forfeited through a prosecution disclaimer.
At summary judgment, this Court found that during reexamination proceedings, Kove clearly disclaimed hierarchical location server configurations and determined that the claims-at-issue for all three patents required location servers in a "non-hierarchical cluster configuration." Kove IO, Inc., 2024 WL 450028, at *5-9. This Court also found that further construction of the term "non-hierarchical" was unnecessary because Kove had "acted as [its] own lexicographer" by describing the meaning of the term in the context of the patents during the reexamination process. Id. at 9 (quoting Thorner v. Sony Computer Ent. Am. LLC, 669 F.3d 1362, 1365 (Fed. Cir. 2012)). The Court concluded that the claim term "plurality of location servers" required “location servers in non-hierarchical structures." Kove IO, Inc., 2024 WL 450028, at *9.
AWS argues that the definition of the term "non-hierarchical" included in the jury instructions "doesn't accurately reflect Kove's disclaimer." Def.'s Mot. for New Trial at 10. AWS fails to acknowledge, however, that the definition of the term "non-hierarchical" included in the jury instructions was adopted, word-for-word, from the description of non-hierarchical structures that Kove provided to the USPTO during the reexamination process. See Def.'s Stmt. of Facts, Ex. 32 at 40. And AWS offers no support for the proposition that it was prejudiced by the construction presented to the jury.
AWS contends that the Court "separated" the non-hierarchical requirements from the "plurality of location servers" claim term. Def.'s Mot. for New Trial at 9. But in the jury instructions, the term "plurality of location servers/plurality of data location servers" was defined as requiring "location servers in non-hierarchical structures." Dkt. no. 876 at 17. Given that the term "non-hierarchical" was included in the "plurality of location servers" definition, AWS's contention that the jury instructions "separated" the two concepts is unpersuasive.
AWS also argues that the Court improperly limited questioning of Dr. Goodrich regarding statements he made during the reexamination process. At trial, AWS attempted to elicit testimony from Dr. Goodrich regarding statements he made to the USPTO about the meaning of the term "non-hierarchical." During the reexamination of the '170 patent, Goodrich submitted a declaration stating that a person of ordinary skill in the art (POSITA) would understand that certain requirements of the claim terms, such as a hash function, mandated a "non-hierarchical clustering approach." Tr. at 824:15825:1. AWS quoted this declaration during its cross-examination of Dr. Goodrich; Kove objected to this questioning as inconsistent with the Court's claim construction. Id. at 825:2-4. At sidebar, AWS argued that its line of questioning was "not claim construction" and instead focused on Dr. Goodrich's contention that "non-hierarchical structures are different than tree structures." Id. at 826:7-9; 826:22-24. This Court sustained Kove's objection. Id. at 827:16-24.
AWS asserts that there was "no substantive reason" to exclude Dr. Goodrich's prior statement about "tree" structures. But the Court sustained the objection under Federal Rule of Evidence 403. Tr. at 827:16-24. Rule 403 provides for the exclusion of evidence "if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence." FED. R. EVID. 403. AWS conceded at trial that Dr. Goodrich's statements to the USPTO were made prior to the Court's claim construction of the term "non-hierarchical." See Tr. at 826:1020. As the Court noted when it sustained Kove's objection, "the problem is that you'd have to go back and recreate -- recreate, you know, what the ground rules were at the time and what was being proposed and what the state of affairs was." Id. at 828:12-19. The jury was instructed to analyze infringement based on the Court's construction of non-hierarchical, which was not in effect at the time of the reexaminations of the Kove patents. The potential for juror confusion and the waste of time required to, in effect, set the scene, substantially outweighed the relatively minimal probative value of Dr. Goodrich's previous statements. There was no error, let alone an error that actually prejudiced AWS.
2. Damages
AWS moves for remittitur, or in the alternative, a new trial to redetermine damages. "A court may vacate a jury's damages award only if it is 'against the clear or great weight of the evidence.'" Sioux Steel Co. v. Prairie Land Mill Wright Servs., No. 16-CV-2212, 2023 WL 5804194, at *8 (N.D. Ill. Sept. 7, 2023) (quoting Unisplay, S.A. v. Am. Elec. Sign Co., 69 F.3d 512, 517 (Fed. Cir. 1995)). In reviewing a jury verdict, this Court must "accord substantial deference to the jury's assessment" of damages. Spina v. Forest Preserve Dist. of Cook Cnty., 207 F.Supp.2d 764, 771 (N.D. Ill. 2002) (citing Ramsey v. Am. Air Filter Co., 772 F.2d 1303, 1313 (7th Cir.1985)). The damages inquiry is confined to three questions: "whether the award is 'monstrously excessive'; whether there is no rational connection between the award and the evidence, indicating that it is merely a product of the jury's fevered imaginings or personal vendettas[;] and whether the award is roughly comparable to awards made in similar cases." Farfaras v. Citizens Bank & Tr. of Chicago, 433 F.3d 558, 566 (7th Cir. 2006) (citation omitted).
AWS challenges the damages award on several bases, many of which this Court considered in ruling on its motion for judgment as a matter of law.
a. Excessive
AWS argues that the damages award is excessive because it "exceeded AWS's revenue from S3 get requests and took nearly all profit from DDB." Def.'s Mot. for New Trial at 13. In support of its argument, AWS cites Shockley v. Arcan, Inc., 248 F.3d 1349 (Fed. Cir. 2001). In Shockley, the defendant argued that the plaintiff's damages expert had not sufficiently supported his opinion regarding projected lost profits. 248 F.3d at 1362. The expert's calculations assumed that, without infringement, the defendant would have sold 80,000 units of the accused product per year over the next fifteen years to a specific buyer. Id. at 1363. But the expert admitted that his projected number of units sold was "an assumption without factual underpinnings" because it was based on "the number of sales that [defendant] told him to assume it would have made." Id. Furthermore, the defendant's salesperson testified that the defendant had attempted to sell many products to the potential buyer the expert referenced for many years with no success. Id. The Federal Circuit vacated the jury's damages award, concluding that the plaintiff's expert had "used a benchmark without any basis in economic reality." Id. The Shockley court also stated that the jury had "assume[d] continued demand and growth rates, profit margins, and other market factors against the clear weight of the evidence." Id. at 1363.
That case is distinguishable, however, because the jury award was based on a reasonable royalty rate rather than future lost profits. "[F]uture lost profits are inherently speculative" and therefore "require a different level of reliable economic evidence" than the factors upon which a reasonable royalty rate is based. Japan Cash Mach. Co. v. MEI, Inc., No. 2:05-CV-01433-RCJ, 2009 WL 10316043, at *34 (D. Nev. Sept. 18, 2009), aff'd, 400 Fed.Appx. 563 (Fed. Cir. 2010) (distinguishing Shockley, 248 F.3d at 1349).
Furthermore, the fact that the damages amount exceeds AWS's revenue for the accused products is insufficient to demonstrate that the award was monstrously excessive. Kove presented evidence that the operating profits for the accused products were not an appropriate measure of total damages. Bergman testified that he believed AWS would agree to pay more the revenue gained from external DDB customers for a license because "internal customers are the significant majority of use on these systems." Tr. at 1125:6-20; 1031:7-24 ("The internal usage far surpasses the amount of external usage. So Amazon itself is using this significantly more than even its third-party customers."). Because AWS's reported operating profits only accounted for external customers, Bergman analyzed the value of internal use of S3 and DDB by assuming that the internal customers derived a similar value from the infringing products as the external customers. Id. at 1034:4-19. Bergman explained how he performed this analysis, which included reviewing how much AWS charges its biggest external customers for use of its products. Id. at 1034:20-1038:23. Additionally, the jury heard testimony that both S3 and DDB drove consumer demand for other AWS products. See id. at 1098:14-1100:23. Taken together, this evidence supports a conclusion that AWS would have agreed to pay more than the operating profits of the accused products in order to obtain a license for the Kove patents.
Furthermore, Bergman based his calculations on actual data rather than projections. He testified that revenue for S3 during the damages period was about $8.64 billion, and $1.1 billion for DDB. Id. at 1032:24-1033:3. Bergman further testified that customer usage of both S3 and DDB grew "exponentially" over time. Id. at 1031:25-1032:10. These numbers were based on financial information provided by Amazon, such as the company's profit-and-loss statements, not pure speculation. Id. at 1032:13-23. This Court cannot conclude that his calculations had no "basis in economic reality." Shockley, 248 F.3d at 1363.
"[E]stimating a reasonable royalty is not an exact science." Summit 6, LLC v. Samsung Elecs. Co., 802 F.3d 1283, 1296 (Fed. Cir. 2015). And the defendant's revenue over the damages period is not the only relevant factor in determining a reasonably royalty. See Lucent Techs., Inc., 580 F.3d at 1324 ("[A]warding damages through litigation attempts to assess 'the difference between [the patentee's] pecuniary condition after the infringement, and what his condition would have been if the infringement had not occurred.'" (quoting Yale Lock Mfg. Co. v. Sargent, 117 U.S. 536, 552 (1886)); Panduit Corp. v. Stahlin Bros. Fibre Works, 575 F.2d 1152, 1159 (6th Cir. 1978) ("Among the relevant facts are: what plaintiff's property was, to what extent defendant has taken it, its usefulness and commercial value as shown by its advantages over other things and by the extent of its use.") (internal quotation omitted). This Court concludes that the jury's verdict is not excessive.
AWS states that due to Kove's failure to meet its evidentiary burden a "remittitur or new trial is warranted." Def.'s Mot. for New Trial at 13. But its remittitur argument is confined to a single sentence stating that "the new trial should be conditional on Kove not accepting a remitted award of $5.25 million-i.e., the maximum of the range provided by Ms. Bennis, which the jury seemed to multiply to reach $525 million." Def.'s Mot. for New Trial at 15. First of all, the contention that the jury applied a 100-times multiplier to Bennis's testimony is wildly speculative; it's far more likely that the jury accepted and adopted the low end of the range argued by the prevailing side's expert, Bergman. That aside, because the Court has concluded that the jury's damages award is not excessive, there is no basis to order a remittitur.
b. Apportionment
AWS asserts that Kove "provided no evidence as to how the undisputed 99.75% of get requests that are served by the request router infringe or could be the basis for any damages" in light of the fact that "[n]one of those requests ever go to the Metadata Nodes, the accused location servers." Def.'s Mot. for New Trial at 13. AWS argues that because only .25% of get requests reach the location servers in DDB, Bergman's damages calculation should have been reduced by 99.75%. This argument is unpersuasive.
To calculate adequate damages for DDB's infringement on the patents, Bergman compared the value of DDB to the value of a hypothetical non-infringing alternative. He testified that his understanding from Dr. Goodrich was that "but for access to these patents, DynamoDB would not have access to the partition map cache." Tr. at 1080:411. He explained that currently the partition map cache resolves get requests 99.75% of the time, and in the non-infringing system, "100% of those requests [would be] going to the metadata nodes." Id. at 1080:12-21. Bergman stated that to compare the value of DDB to the non-infringing alternative, he reviewed an AWS document demonstrating that it had performed a "quantifiable analysis to consider what the cost impact would be by taking out that partition map cache." Id. at 1080:22-1081:1. AWS's analysis demonstrated that removing the partition map cache would result in between $571 million and nearly $700 million in costs in order to "add additional equipment to be able to support the operations that that cache has," which Bergman stated would result in up to $1 billion in extra costs over the entire damages period. Id. at 1080:2-17.
AWS's assertion that the 99.75% of get requests that are resolved through the partition map cache cannot be the basis for a damages award ignores Kove's evidence that the functionality of the partition map cache depends in part on the practice of the claims of the Kove patents. Dr. Goodrich testified that request routers are only able to store data locations through the use of a hash function. He stated that without the hashing claims of the patents-in-suit, the partition map cache would contain "hundreds of terabytes of data" and "each request router would store only .007 percent of all the location data," which would require "that effectively all the get requests would have to go to the metadata nodes" rather than the partition map cache. Id. at 738:22-739:10. He also quoted from an Amazon document detailing its analysis of the consequences of having all get requests sent to the metadata nodes. The document stated that "[t]he metadata hardware footprint would have to scale up 1500 times to handle an L-1 partition map cache failure today." Id. at 739:22-740:9. Dr. Goodrich explained that a partition map cache failure would prevent access to the partition map cache, and as a result "you have to blow up the metadata node fleet like crazy." Id. at 740:10-15.
"As a substantive matter, it is the 'value of what was taken' that measures a 'reasonable royalty' under 35 U.S.C. § 284." Power Integrations, Inc. v. Fairchild Semiconductor Int'l, Inc., 904 F.3d 965, 977 (Fed. Cir. 2018) (citation omitted). The expert testimony and Amazon documents Kove presented at trial indicate that a substantial part of the "value" of the Kove patents to DDB is the ability to use a hash function in order to support the partition map cache. Bergman confirmed at trial that he held this opinion despite understanding that the request routers, which contain the partition map cache, are not the component of DDB that Kove alleged satisfied the patents' "location server" requirement. See Tr. at 1205:3-12. This opinion is not against the manifest weight of the evidence because in the non-infringing alternative Bergman developed, none of the get requests would be sent to the partition map cache, and all get requests would be sent to the metadata nodes. Therefore, Bergman's damages calculation evaluated the impact of requiring every get request to be sent to the metadata nodes, not just the value of the .25% of requests that reach the metadata nodes.
Furthermore, on both direct and cross-examination, Bergman was asked why he didn't reduce his proposed damages award for DDB by 99.75% to reflect the number of get requests that do not reach the metadata nodes. On cross- examination he stated that he did not agree that calculating the damages award based on the proportion of get requests that reach the metadata nodes was an "appropriate way to do the analysis." Id. at 1205:25-1206:8; see id. at 1081:20-1082:2. Bergman's apportionment opinions were supported by Kove's evidence regarding the functionality of the accused technology and therefore are distinct from the opinions that the Federal Circuit has found to have been "plucked out of thin air." LaserDynamics, Inc. v. Quanta Computer, Inc., 694 F.3d 51, 69 (Fed. Cir. 2012). Bergman's apportionment opinions related to DDB were not against the manifest weight of the evidence.
AWS also seeks a new trial based on Bergman's apportionment of multiple features of S3. As the Court discussed in denying AWS's motion for judgment as a matter of law on the same issue, Bergman stated multiple times at trial that he relied on Kove's other expert witnesses to gain an understanding of the functionality of the accused products and the benefits the Kove patents provided to those products. See, e.g., Tr. at 1186:17-24 (explaining that his understanding about how KFCs and BMs transfer iNodes came from Dr. Goodrich). This reliance was also the basis for Bergman's consideration of each of the functions of S3 that AWS identifies. See Tr. at 1047:17-1048:7 ("[I]t's my understanding that S3 uses hashes and hash tables in the KFC fleet to locate cached iNodes, which is what contains location information for objects."); 1064:24-1065:4 ("It's my understanding from Dr. Goodrich that the scaling claims provide the ability for S3 to transfer location information between location servers when one of those servers reaches a predetermined performance limit."); 1186:17-23 ("That's my understanding based on my . . . based on Dr. Goodrich."). AWS does not point to any evidence that calls into question Bergman's calculations and simply cites multiple portions of Bergman's testimony that it asserts demonstrates Bergman's flawed understanding of the accused products. Def.'s Mot. for New Trial at 14. This is insufficient to demonstrate that Bergman's apportionment conclusions were against the manifest weight of the evidence.
AWS is not entitled to a new trial based on Kove's apportionment evidence.
c. Goldilocks
AWS argues that Bergman's testimony related to Amazon's "Goldilocks" product was inadequate for the same reasons it argues there was no legally sufficient basis for his testimony. As discussed above, this argument lacks merit. Bergman explained that he relied on Dr. Goodrich and Fanous to understand the features of each product. Tr. at 1199:22-1200:12. He applied a 20% price premium to AWS's Goldilocks over Google's Nearline despite differences between the two products based on this understanding as well as AWS documents comparing the two products. At trial Bergman discussed a document outlining AWS's internal valuation of its product over Nearline. Tr. at 1071:3-15. Bergman also reviewed multiple AWS documents identifying seamless tiering as a key distinction between Goldilocks and Nearline. Id. at 1070:10-1071:2. Bergman provided ample support for his opinion that seamless tiering was a core distinction between the two products that accounted for at least part of the price difference.
AWS notes that one of its witnesses testified about features of the products that it believes Bergman failed to consider. See Def.'s Mot. for New Trial at 15. The fact that AWS offered competing expert testimony is insufficient to demonstrate that the jury's verdict was irrational or not factually supported. And even if this Court were to find AWS's expert more persuasive than Kove's, this Court cannot grant a new trial "just because it believes the jury got it wrong." Whitehead v. Bond, 680 F.3d 919, 928 (7th Cir. 2012). AWS has offered no evidence that Bergman's Goldilocks testimony, individually or in connection with Bergman's other alleged errors, demonstrates that the jury verdict was against the manifest weight of the evidence.
d. Bargaining split
This Court has concluded that Bergman's bargaining split opinions did not provide a basis for granting judgment as a matter of law on damages. For similar reasons, Bergman's analysis does not support granting AWS's motion for a new trial. Bergman explained to the jury the quantified and unquantified benefits that AWS received from use of Kove's patents, the impact of the use of the Kove patents on the growth of other AWS services and the significant resulting benefits to AWS, and AWS's typical business practices with respect to licensing agreements. Tr. at 1097:6-1104:4. As noted above, Bergman's analysis was rooted in the facts of the case, not on "speculation or guesswork." Lucent Techs., Inc., 580 F.3d at 1335.
The jury had ample credible evidence supporting its decision on damages, and its damages award was not against the manifest weight of the evidence. This Court therefore denies AWS's motion for a new trial on damages.
C. Kove's motion to amend the judgment and for attorneys' fees
1. Pre- and post-judgment interest
Kove requests an award of prejudgment interest and postjudgment interest on all damages incurred prior to and after the Court's entry of judgment. An award of prejudgment interest under 35 U.S.C. § 284 "serves to make the patent owner whole, since his damages consist not only of the value of the royalty payments but also of the forgone use of the money between the time of infringement and the date of the judgment." Gen. Motors Corp. v. Devex Corp., 461 U.S. 648, 656 (1983). The Supreme Court has held that "prejudgment interest should ordinarily be awarded absent some justification for withholding such an award." Id. at 657.
AWS argues that this Court should not award prejudgment interest because Kove is responsible for delaying the case's resolution. The parties also disagree about the proper calculation for a prejudgment interest award in this case.
a. Delay
A court may limit or deny prejudgment interest "where the patent owner has been responsible for undue delay in prosecuting the lawsuit." Gen. Motors Corp., 461 U.S. at 657. But "[a]bsent prejudice to the defendants, any delay by [the patentee] does not support the denial of prejudgment interest." Crystal Semiconductor Corp. v. TriTech Microelectronics Int'l, Inc., 246 F.3d 1336, 1362 (Fed. Cir. 2001).
AWS argues that Kove unduly delayed prosecuting its claims by waiting to bring suit for nearly four years after it began considering a suit against AWS for patent infringement. Dr. John Overton, one of the inventors of the Kove patents, testified that he began discussing potential patent infringement litigation with attorneys sometime in 2014 or 2015 and that some of those discussions were about a potential lawsuit against AWS. Tr. at 368:23-369:4; 372:7-16. He also testified that the patents' co-inventor, Dr. Steph Bailey, analyzed various Amazon products while researching potential infringement cases based on the patents. Id. at 381:24-382:9. Additionally, Dr. Overton stated that throughout 2017 and 2018 Kove had communications with AWS about the potential sale of a Kove memory product unrelated to the patents-in-suit. Id. at 373:21374:9. When asked if he thought it would have been "fair" to inform AWS of potential infringement during these communications, Overton responded:
This is a complicated thing. You're trying to -- we're a small company. Amazon is the largest company in the world. You do your best to make the best judgments possible on how to advance the interests that are mutually beneficial, so you do your best. And it did not make any sense to say to them that we were considering the lawsuit, and if something else had happened and we had done business, maybe there would have been different outcomes. I can't speculate all the ways that one would evaluate something like this.
But we were considering at this time, yes, that they were infringing or we believed so, whatever that is, we were considering the lawsuit, yes, and we were also considering that we could do business. You know, the specifics of that, it's very dynamic.Id. at. 376:15-377:6. Kove did not attempt to inform AWS of its patents or any potential infringement until it filed the present suit in December 2018.
AWS argues that Kove's conduct is similar to the actions of the plaintiff in Crystal Semiconductor Corp. In that case, the defendant adduced evidence that the plaintiff unduly delayed filing via the testimony of the plaintiff's former president, who stated that the company had sent letters to over thirty companies notifying them of potential infringement but intentionally excluded the defendants. Crystal Semiconductor Corp., 246 F.3d at 1362. The same witness also explained that the plaintiff did not file suit because it wanted to establish a "business relationship" with a company that had a relationship with the defendants. Id. The Federal Circuit found that the plaintiff's "selfserving" delay in initiating the suit was a "litigation tactic." Id. The court also concluded that the delay "caused the damages owed by [defendants] to escalate" and "resulted in prejudice to the defendants." Id. The Federal Circuit affirmed the district court's denial of prejudgment interest. Id.
Kove argues that AWS's reliance on Crystal is misplaced because that case is "irreconcilable" with the Supreme Court's decision in SCA Hygiene Products Aktiebolag v. First Quality Baby Products, LLC, 580 U.S. 328 (2017). In that case, the Supreme Court held that a defendant could not assert a laches defense against a patent infringement claim that was brought within the damages period outlined in 35 U.S.C. § 286. Id. at 346. Kove does not cite, nor is this Court aware of, any binding or persuasive authority indicating this bar on laches as an affirmative defense to liability also precludes denial or limitation of prejudgment interest on a damage award once liability is established. See Milwaukee Elec. Tool Corp., 288 F.Supp.3d at 907 ("Plaintiffs do not explain the connection between laches and Section 284, and the Court is not satisfied that they have much to do with each other."). And since the decision in SCA Hygiene Products Aktiebolag, the Federal Circuit has continued to cite Crystal Semiconductor Corp. and apply the undue delay standard in determining issues relating to prejudgment interest. See Kaufman v. Microsoft Corp., 34 F.4th 1360, 1375 (Fed. Cir. 2022) (describing how the facts of the case were distinct from those present in Crystal Semiconductor Corp.).
The evidence in the record supports a finding that Kove unduly delayed in bringing suit such that prejudgment interest should be limited. It is undisputed that Kove was aware of a possible claim for infringement against AWS at least four years before it filed its complaint in this case. District courts have considered whether the plaintiff has offered a justification for its delay in filing suit when determining whether to award prejudgment interest. See, e.g., Kaneka Corp. v. SKC Kolon PI, Inc., 198 F.Supp.3d 1089, 1124 (C.D. Cal. 2016) ("[Plaintiff] offers no compelling explanation for why, after it was aware of a possible claim for infringement, it waited four years to file suit."); Saint-Gobain Autover USA, Inc. v. Xinyi Glass N. Am., Inc., 707 F.Supp.2d 737, 767 (N.D. Ohio 2010) ("Plaintiffs fail to offer any justification (or even explanation) for this substantial delay."). Aside from its efforts to build a business relationship with the defendant, which the Federal Circuit rejected as a legitimate reason for delay in Crystal Semiconductor Corp., Kove has offered no explanation for why it waited until 2018 to file suit despite knowing as early as 2014 that AWS's products potentially infringed on its patents.
AWS has also adduced evidence that it was prejudiced by Kove's undue delay. The Federal Circuit has stated that a defendant's bare assertion that it could have modified its products to avoid infringement is insufficient to demonstrate prejudice. Kaufman, 34 F.4th at 1375. Here, AWS offers more than a general statement that it could have implemented noninfringing alternatives. Scott Hayden, the president of Amazon Technologies, Inc., testified that the company "respect[s] the intellectual property rights of others," and he stated that upon notification that it may be infringing AWS would "enter a negotiation or we would stop doing what we were doing." Tr. at 989:8-13. Hayden further testified that although AWS did not believe that it infringed on the Kove patents, it still would have been willing to pay for a license to practice those patents because the company has "taken licenses to things that we don't infringe because it helps us avoid the cost of litigation." Tr. at 991:12-19. Based on the evidence regarding AWS's business practices regarding potentially infringing products, it is reasonable to conclude that AWS was prejudiced by the delay.
That said, the Court is not persuaded that this delay warrants denying prejudgment interest in full. Awarding prejudgment interest is "the rule, not the exception." Crystal Semiconductor Corp., 246 F.3d at 1361. And though Overton's testimony indicates that the desire to establish a business relationship impacted the timing of the lawsuit, unlike the defendant in Crystal, AWS has not adduced evidence that Kove's delay was a litigation tactic or that Kove was selectively enforcing its patent rights. The Court will therefore limit prejudgment interest to the period between Kove's filing of the complaint to the date of the judgment. See Milwaukee Elec. Tool Corp., 288 F.Supp.3d at 907 (awarding prejudgment interest "only from the institution of suit until the entry of judgment.").
Kove also argues that, because the Court disposed of AWS's equitable estoppel and waiver defenses at summary judgement, AWS cannot rely on evidence regarding "the same alleged delay based on the same circumstances" to limit prejudgment interest. Pl.'s Mot. to Amend the J. Reply at 3. But the standard for establishing material prejudice as required to prevail on an equitable estoppel or waiver defense is distinct from the standard for proving prejudice due to an undue litigation delay. Specifically, to demonstrate material prejudice, a defendant typically has to demonstrate "a change of economic position or loss of evidence" resulting from the plaintiff's conduct. A.C. Aukerman Co. v. R.L. Chaides Const. Co., 960 F.2d 1020, 1043 (Fed. Cir. 1992), abrogated on other grounds by SCA Hygiene Prods. Aktiebolag, 580 U.S. 328. In contrast, to demonstrate prejudice due to undue delay, a defendant need only show that it "would have mitigated its damages but for the delay." MHL Custom, Inc. v. Waydoo USA, Inc., No. CV 21-0091-RGA, 2023 WL 5805889, at *7 (D. Del. Sept. 7, 2023).
The Court concludes that Kove is entitled to prejudgment interest for the period from the filing of its complaint through the entry of judgment.
b. Calculation
Kove argues that it is entitled to prejudgment interest calculated at the prime rate, compounded monthly. AWS, on the other hand, argues that the prejudgment interest award should be compounded quarterly at the 52-week Treasury Bill (T-Bill) rate. Regional circuit law is the controlling authority on calculation of prejudgment interest. See Transmatic, Inc. v. Gulton Indus., Inc., 180 F.3d 1343, 1347-48 (Fed. Cir. 1999).
i. Rate
In this circuit, the prime rate is the preferred rate for prejudgment interest awards. See Gorenstein Enters., Inc. v. Quality Care-USA, Inc., 874 F.2d 431, 436 (7th Cir. 1989) ("[W]e suggest that district judges use the prime rate for fixing prejudgment interest where there is no statutory interest rate."); First Nat. Bank of Chi. v. Standard Bank & Tr., 172 F.3d 472, 480 (7th Cir. 1999) ("We hold today that to . . . award something other than the prime rate is an abuse of discretion, unless the district court engages in such a refined calculation."). The Seventh Circuit has required district courts to use the prime rate in the absence of a "statutorily defined rate" or a process of "refined rate-setting" that can be used to determine a "more accurate market rate for interest." Frey v. Coleman, 903 F.3d 671, 682 (7th Cir. 2018) (internal quotation omitted).
The majority of courts in this Circuit have followed this mandate. See, e.g., Raffel Sys., LLC v. Man Wah Holdings LTD, Inc., No. 18 CV 1765, 2023 WL 3375853, at *23 (E.D. Wis. May 11, 2023) ("I see no reason to depart from the general practice in the circuit of using the prime rate to calculate prejudgment interest."); ABS Glob., Inc. v. Inguran, LLC, No. 14 C 503, 2020 WL 2405380, at *8 (W.D. Wis. May 12, 2020) ("[C]onsistent with the Federal and Seventh Circuit's practice, this court generally awards pre-judgment interest at the prime rate."); Ameritox, Ltd. v. Millennium Health, LLC, No. 13 C 832, 2016 WL 9460661, at *3 (W.D. Wis. July 18, 2016) ("[T]he court will follow the practice approved by the Federal Circuit and Seventh Circuit, which is also consistent with its own practice, by awarding prejudgment interest based on the prime rate."); see also Aero Prod. Int'l, Inc. v. Intex Recreation Corp., No. 02 C 2590, 2004 WL 2091996, at *2 (N.D. Ill. Sept. 16, 2004), aff'd, 466 F.3d 1000 (Fed. Cir. 2006) ("[T]he Treasury Bill rate would not adequately compensate Plaintiffs.). AWS points to a small number of cases where courts in this district have departed from this approach. See Black & Decker Inc. v. Robert Bosch Tool Corp., No. 04 C 7955, 2006 WL 3359349, at *11 (N.D. Ill. Nov. 20, 2006), vacated, 260 Fed.Appx. 284 (Fed. Cir. 2008). But AWS has not produced any evidence to support the application of any rate other than the prime rate. See Milwaukee Elec. Tool Corp., 288 F.Supp.3d at 908 (E.D. Wis. 2017) ("The parties have not supplied authorities, evidence, or argument that equips the Court to engage in the nuanced rate-setting analysis needed to depart from the prime rate.").
The Court concludes that the prime rate is the appropriate prejudgment interest rate.
ii. Compounding
The parties dispute whether interest should be compounded monthly or quarterly. During discovery, AWS produced three running royalty license agreements with other businesses that included a commitment to make payments on a quarterly basis. See Def.'s Resp. to Pl.'s Mot. to Amend the J., Ex. 6 at AMZKOVE000088426; Ex. 7 at AMZKOVE000089327; Ex. 8 at AMZKOVE000087899. AWS's past license agreements containing a quarterly payment schedule weigh in favor of awarding prejudgment interest compounded on a quarterly basis as well. See Trading Techs. Int'l, Inc. v. IBG LLC, No. 10 C 715, 2022 WL 103894, at *3 (N.D. Ill. Jan. 11, 2022) (applying quarterly compounding based on defendant's business practices with other licensees); Kolcraft Enters., Inc. v. Chicco USA, Inc., No. 09 C 3339, 2019 WL 4242482, at *19 (N.D. Ill. Sept. 6, 2019), aff'd sub nom. Kolcraft Enters., Inc. v. Artsana USA, Inc., 819 Fed.Appx. 939 (Fed. Cir. 2020) (same).
Kove argues that its proposal to compound interest on a monthly basis better "reflects reality," but it does not connect its proposal to the facts of this case and cites only non-patent cases in support of its argument. See Pl.'s Mot. to Amend the J. Reply at 6-8. Other courts in this district have concluded that monthly compounding is an appropriate standard outside the context of a patent infringement case. See Pl.'s Mot. to Amend the J. Reply at 7 n. 3 (listing cases). But quarterly compounding has been determined to be an appropriate standard for interest in patent infringement suits. Milwaukee Elec. Tool Corp., 288 F.Supp.3d at 909 (identifying quarterly compounding as the "ordinary approach to interest calculation"); Bayer Healthcare LLC v. Baxalta Inc., No. 16-CV-1122-RGA, 2019 WL 4016235, at *7 (D. Del. Aug. 26, 2019), aff'd, 989 F.3d 964 (Fed. Cir. 2021) ("[I]t seems that standard industry practice includes interest . . . compounded quarterly.").
This Court concludes that Kove is entitled to prejudgment interest at the prime rate, compounded quarterly, for the period following Kove's filing of the lawsuit. AWS's damages expert Bennis calculated prejudgment interest for the period from December 12, 2018 through April 10, 2014 using the prime rate and quarterly compounding. She concluded that the correct amount using these parameters is $147,706,733. The parties are directed to update this calculation and are to file a joint statement on the correct amount by August 30, 2024.
AWS does not challenge the award of postjudgment interest in this case, and indeed, postjudgment interest is provided for by statute. Kove is entitled to postjudgment interest under 28 U.S.C. § 1961.
The applicable postjudgment interest rate is "equal to the weekly average 1-year constant maturity Treasury yield . . . for the calendar week preceding the date of the judgment." 28 U.S.C. § 1961(b). The parties agree that the applicable rate for the week of April 1, 2024 was 5.04%. Bennis Decl. ¶ 7.
2. Attorneys' fees
Under 35 U.S.C. § 285, "[t]he court in exceptional cases may award reasonable attorney fees to the prevailing party" in a patent infringement lawsuit. An exceptional case is "one that stands out from others with respect to the substantive strength of a party's litigating position (considering both the governing law and the facts of the case) or the unreasonable manner in which the case was litigated." Octane Fitness, LLC v. ICON Health & Fitness, Inc., 572 U.S. 545, 554 (2014). Determining whether a case is considered "exceptional" requires "considering the totality of the circumstances," as "[t]here is no precise rule or formula for making these determinations." Id. A district court may consider factors such as "frivolousness, motivation, objective unreasonableness of a case's factual or legal components, and the need in particular circumstances to advance considerations of compensation and deterrence." Univ. of Utah v. Max-Planck-Gesellschaft zur Foerderung der Wissenschaften e.V., 851 F.3d 1317, 1322 (Fed. Cir. 2017). An award of attorneys' fees is "limited to circumstances in which it is necessary to prevent a 'gross injustice' or bad faith litigation." Microsoft Corp. v. WebXchange Inc., 715 F.Supp.2d 598, 603 (D. Del. 2010) (quoting Forest Labs., Inc. v. Abbott Labs., 339 F.3d 1324, 1329 (Fed. Cir. 2003)). A party must prove entitlement to attorneys' fees by a preponderance of the evidence. Octane Fitness, 572 U.S. at 557.
Kove argues that this case is exceptional because AWS engaged in various instances of misconduct, including dropping its affirmative defenses on the eve of trial, introducing new infringement theories at trial, and violating various pretrial rulings.
1. Affirmative defenses
AWS originally asserted twelve affirmative defenses against Kove. This Court ruled in Kove's favor on several of the defenses at summary judgment. Kove IO, Inc. v. Amazon Web Servs., Inc., 2024 WL 450028, at *18-30. Shortly before the final pretrial conference, AWS dropped its inequitable conduct and double patenting defenses. And five days before opening arguments, AWS dropped its remaining defenses. AWS contends that it declined to pursue these defenses at trial for "strategic reasons." Def.'s Resp. to Pl.'s Mot. for Fees at 12. Kove asserts that AWS attempted to secure a tactical advantage by "driving up litigation costs, delaying trial, and distracting from the matters legitimately in dispute." Pl.'s Mot. for Fees at 13.
In support of its argument, Kove cites GoTV Streaming, LLC v. Netflix, Inc., No. 2:22-CV-07556-RGK-SHK, 2024 WL 1832389 (C.D. Cal. Mar. 26, 2024). That case is distinguishable. There the court considered whether the defendant's abandonment of its invalidity defenses shortly after the trial began made the case exceptional. Although the defendant insisted that it dropped its defenses as part of a reasonable trial strategy, the plaintiff contended that the defendant "feigned pursuit of the defense to distract Plaintiff and waste its resources." Id. at *7. In determining that the defendant's conduct merited a fee award, the court noted that the plaintiff had presented "several instances of suspicious conduct," including the defendant's disinterest in streamlining its invalidity defenses and the lack of attention paid to invalidity during the defendant's opening statement. Id. at *8.
Here, Kove points to no evidence to support its contention that AWS's conduct was motivated by malicious intent rather than ordinary trial strategy. And Kove cites no legal authority demonstrating that the pursuit of affirmative defenses alone, in the absence of evidence that those defenses "lacked substantive strength," is sufficient to render a case exceptional. SRI Int'l, Inc. v. Cisco Sys., Inc., No. CV 13-1534-RGA, 2020 WL 1285915, at *4 (D. Del. Mar. 18, 2020), aff'd in part, rev'd in part on other grounds, 14 F.4th 1323 (Fed. Cir. 2021) (noting that invalidity defenses that the defendant dropped shortly before trial were weak); Tinnus Enters., LLC v. Telebrands Corp., 369 F.Supp.3d 704, 720 (E.D. Tex. 2019) ("[Defendant] knew its inequitable conduct claim was weak because it barely survived Plaintiffs' initial motion to dismiss."); EagleView Techs., Inc. v. Xactware Sols., Inc., 522 F.Supp.3d 40, 61-62 (D.N.J. 2021) (describing the defendant's attempts to re-litigate defenses that were previously rejected by the court). In the absence of evidence that Kove acted with an improper motive, this Court concludes that AWS's abandonment of its invalidity defenses did not rise to the level of "exceptional" conduct.
2. New infringement theories
Kove argues that AWS attempted to raise new non-infringement arguments for the first time at trial. In its opening statement, AWS differentiated S3 and DDB from the Kove patents by emphasizing the accused products' "security" features and "one-key system." Specifically, AWS stated, "[s]ecurity, security, security, ladies and gentlemen. That's what defines the decision choices that we're going to hear about that separates the patents from S3 and the patents from DDB." Tr. at 226:9-13. AWS also noted that the patents teach a "one-key system" with "no security," Tr. at 216:8-10, and contrasted that system with S3's two-key system and DDB's three-key system. Id. at 218:8-14; 222:11-16. The Court later issued an instruction clarifying that AWS offered testimony regarding AWS's security features and multi-key systems "to explain how its products work and why it developed those products the way it did," and that this testimony was "not to be considered in determining whether AWS's accused products infringe Kove's patents." Id. at 1352:4-1353:2.
It is unclear whether Kove is arguing that AWS (1) improperly pursued a meritless non-infringement position or (2) engaged in bad-faith litigation conduct by introducing a previously undisclosed non-infringement theory at trial. If Kove's argument is the former, it is unpersuasive. To determine whether a case is exceptional under section 285, the court examines the "substantive strength of the party's litigating position," not the "correctness or eventual success of that position." SFA Sys., LLC v. Newegg Inc., 793 F.3d 1344, 1348 (Fed. Cir. 2015). During the Court's consideration of the corrective instruction, AWS stated that the purpose of its testimony was to highlight that security, rather than other benefits such as scaling, was the primary factor motivating the company's design decisions. Tr. at 1235:10-1236:6. Based on AWS's explanation, the Court removed from the curative instruction language indicating that these design choices had "no bearing" on the jury's infringement analysis, recognizing that the decisions that underline the design of a product can be relevant to infringement. AWS's position that the decision choices for S3 and DDB bore some relation to infringement was not frivolous or objectively unreasonable. See Greatbatch Ltd. v. AVX Corp., No. CV 13-723-LPS, 2017 WL 3085055, at *3 (D. Del. July 20, 2017) (ruling that design decisions for allegedly infringing product were probative of whether infringement occurred). Although AWS's position was ultimately a losing one, it was not "so substantively weak as to render this case exceptional." Ferring Pharms. Inc. v. Par Pharm., Inc., No. 1:15-CV-00173-RGA, 2018 WL 6696040, at *1 (D. Del. Dec. 19, 2018).
If Kove's argument is the latter, it likewise lacks merit. During its opening statement, AWS noted that its witnesses would share the security features of the accused products and explain why security is vital to their functionality. Tr. at 219:1-21. AWS later mentioned the accused products' security and multi-key system features in the context of its description of the "decision choices" that motivated the products' design. Tr. at 226:9-13. This Court issued a corrective instruction to clarify the distinction between testimony describing AWS products' features and testimony tying those features to the parties' infringement contentions. But AWS's actual theories of non-infringement focused on the language of the claim limitations, not security or multikey systems. And as Kove notes, "the core dispute at trial was not how the products worked, but whether that functionality infringed." Pl.'s Mot. for Fees Reply at 14. Following this Court's instruction, AWS continued to reference security and multi-key systems in the context of the design choices for the accused products rather than as direct evidence of non-infringement. See e.g., Tr. at 1340:2-5. Kove had the opportunity to object to these questions, and it chose not to do so. Id. at 1339:1-5. Kove also had the opportunity to question witnesses on cross-examination about these design choices examination. Based on these facts, the Court finds that AWS did not pursue the litigation in an unreasonable manner by its reference to security and multikey system features.
Kove also asserts that AWS "misstated the law" by implying that the jury could find non-infringement if it concluded that the accused products were "different" from the patents. Pl.'s Mot. for Fees at 7. Specifically, during its opening statement, AWS stated that the "test for infringement or non-infringement" is whether the accused products are "doing the same thing or something different than the patents, same or different." Tr. at 209: 209:24-210:5. AWS repeated this "same or different" language throughout its opening statement. See, e.g., Id. at 223:3-9; 227:13-14. But AWS did not ask any of its expert witness to provide their infringement analysis using this "same or different" language, and in its closing statement AWS expressly stated that same or different is "not the legal standard." Id. at 1950:8-14. There is no basis for a finding that AWS's "same or different" phrase was meant to confuse the jury with an incorrect recitation of patent law.
3. Pretrial ruling violations
Kove asserts that AWS's trial conduct included "serial attempts to undo the Court's claim construction rulings." Pl.'s Mot. for Fees at 8. Kove notes that AWS continuously raised claim construction arguments prior to the trial. This case involved numerous complicated issues, and it included a number of claim construction disputes. Because "district courts may engage in rolling claim construction," AWS's attempts to resolve perceived disputes over the language of the claim limitations cannot be considered "exceptional." Pfizer, Inc. v. Teva Pharms., USA, Inc., 429 F.3d 1364, 1377 (Fed. Cir. 2005) (citation omitted). AWS's claim construction requests were not so unreasonable as to warrant the shifting of attorneys' fees.
Additionally, Kove contends that AWS litigated this case in an unreasonable manner by repeatedly violating this Court's claim construction and motion in limine rulings. In support of its argument, Kove cites Tinnus Enterprises, LLC. But the defendants in that case engaged in a far more egregious pattern of litigation misconduct than what is apparent here, including the filing of "numerous unmeritorious motions" and "sanctionable discovery misconduct." Tinnus Enters., LLC, 369 F.Supp.3d at 745. The court in Tinnus Enterprises, LLC also noted that the defendants "flagrantly and continuously violated motions in limine and sustained objections." Id. But the conduct at issue there involved the introduction of inadmissible evidence through witness questioning and the continued solicitation of the same prohibited testimony from the same witness after the court sustained multiple objections. Id. at 720-722.
Although this Court sustained at trial multiple objections relating to in limine rulings, there is no indication that AWS lacked a good faith belief that its questioning was in compliance with the Court's previous rulings. And rather than constantly attempting to elicit prohibited testimony as the defendants did in Tinnus Enterprises, LLC, AWS generally rephrased its questions following objections, Tr. at 835:1-8; moved on to a new topic, Id. at 1463:17-1464:7; or withdrew the question. Id. at 836:1-6. And none of AWS's arguments involved "positions contrary to its own evidence or prior statements," which is generally found in cases imposing attorney's fees based on litigation conduct. BroadSoft, Inc. v. CallWave Commc'ns, LLC, No. CV 13-711-RGA, 2019 WL 3750817, at *6 (D. Del. Aug. 8, 2019). Furthermore, the court in Tinnus Enterprises, LLC noted that the jury's willfulness finding "further support[ed] a finding that this case is exceptional." Tinnus Enters., 369 F.Supp.3d at 746. Here, there is no willfulness finding to further support Kove's exceptionality argument. See LG Display Co. v. AU Optronics Corp., 722 F.Supp.2d 466, 474 (D. Del. 2010).
Finally, Kove asserts that "AWS's repeated flouting of the Court's rulings necessitated a corrective instruction." Pl.'s Mot. for Fees at 13. On the contrary, the Court's corrective instructions with respect to claim construction were issued to "avoid confusion," not penalize AWS for its conduct. Tr. at 1876:22-1877:3.
Based on the totality of the circumstances, this Court denies Kove's motion to declare this case exceptional and for an award of attorneys' fees.
Conclusion
For the foregoing reasons, the Court denies AWS's motions for judgment as a matter of law [dkt. nos. 870, 891, 901] and its motion for a new trial [dkt. no. 902]. The Court grants in part Kove's motion to amend the judgment and for attorneys' fees [dkt. no. 909] with respect to the pre- and post-judgment interest, but otherwise denies that motion. The parties' joint statement regarding prejudgment interest as described in this order is to be filed by no later than August 27, 2024.