Uniram Technology, Inc. v. Taiwan Semiconductor Manufacturing Co.

(2007)

UNIRAM TECHNOLOGY, INC., Plaintiff,
v.
TAIWAN SEMICONDUCTOR MANUFACTURING COMPANY, Defendant.

No. C-04-1268 VRW.

United States District Court, N.D. California.

September 5, 2007.

ORDER

VAUGHN R. WALKER, Chief Judge.

This is a trade secret misappropriation dispute between plaintiff UniRAM Technology, Inc ("UniRAM") and defendants Taiwan Semiconductor Manufacturing Company LTD and TSMC North America (collectively "TSMC"). In its original complaint, UniRAM also sued Monolithic Systems Technology, Inc ("MoSys"), but that defendant has since settled. TSMC now seeks summary judgment that (1) UniRAM's claims are barred by the statute of limitations and (2) TSMC did not misappropriate UniRAM's trade secrets. For the reasons that follow, TSMC's motion is GRANTED in part and DENIED in part.

I

The trade secrets at issue concern a new method of manufacturing dynamic random access memory ("DRAM") invented by Dr. Jeng-Jye Shau. DRAM is a type of memory that stores data within a circuit. The DRAM memory cells must be embedded into the chip during the manufacturing process. One method of doing this is the EmbDRAM process, while another method is called a logic process. Shau claims that embedding the DRAM by using the logic process was one of his ideas. In 1996, Shau approached TSMC, a foundry of computer chips and circuits, to manufacture his new circuits. Shau decl. at ¶ 5. As part of the process, Shau discussed the invention with TSMC and disclosed "tape outs" of his invention, which included different "features" for manufacturing his device. Id. at ¶¶ 11, 17. Among these features are circuit architecture characteristics such as "EDRAM macro," "small block," "hidden refresh," "SRAM interface," "planar capacitors" and "standard logic" to name a few. Doc. # 375 at 8 (chart). UniRAM contends its DRAM trade secrets consist of particular combinations of these features. UniRAM asserts that Shau disclosed to TSMC a total of twelve DRAM trade secrets and that TSMC transferred these secrets to MoSys and to Matsushita, which then began to produce similar products.

TSMC attacks the merits of the misappropriation claim. TSMC disputes that it ever acquired knowledge of UniRAM's trade secrets, asserting that the information in the tapeouts was insufficient as a matter of law to constitute disclosure of trade secrets. Id at 4. TSMC contends the tapeouts presented only hundreds of thousands of possible combinations of features, and TSMC could not have known the specific combinations which UniRAM claims are secret. Id at 6. TSMC then argues that it never communicated any alleged secrets to MoSys or Matsushita and accordingly did not misappropriate trade secrets as a matter of law. Id at 10. TSMC further contends that a number of allegedly infringing products do not use one of the trade secret elements at issue. Id. at 11. Lastly, TSMC claims that UniRAM knew or should have known of any alleged misappropriation prior to a time within the statute of limitations period. The court addresses these contentions in turn.

II

In reviewing a summary judgment motion, the court must determine whether genuine issues of material fact exist, resolving any doubt in favor of the nonmoving party. "[S]ummary judgment will not lie if the dispute about a material fact is ``genuine,' that is, if the evidence is such that a reasonable jury could return a verdict for the nonmoving party." Anderson v. Liberty Lobby, 477 U.S. 242, 248, 106 S. Ct. 2505, 91 L. Ed. 2d 202 (1986). "Only disputes over facts that might affect the outcome of the suit under the governing law will properly preclude the entry of summary judgment." Id. The burden of establishing the absence of a genuine issue of material fact lies with the moving party. Celotex Corp. v. Catrett, 477 U.S. 317, 322-23, 106 S. Ct. 2548, 91 L. Ed. 2d 265 (1986). Summary judgment is granted only if the moving party is entitled to judgment as a matter of law. FRCP 56(c).

The nonmoving party may not simply rely on the pleadings, however, but must produce significant probative evidence, by affidavit or as otherwise provided in FRCP 56, supporting the claim that a genuine issue of material fact exists. TW Elec. Serv. v. Pacific Elec. Contractors Ass'n, 809 F.2d 626, 630 (9th Cir.1987). Conclusory, speculative testimony in affidavits and moving papers is insufficient to raise genuine issues of fact and defeat summary judgment. Thornhill Publishing Co., Inc. v. GTE Corp., 594 F.2d 730, 738 (9th Cir. 1979). The evidence presented by the nonmoving party "is to be believed, and all justifiable inferences are to be drawn in his favor." Anderson, 477 U.S. at 255, 106 S. Ct. 2505. "[T]he judge's function is not himself to weigh the evidence and determine the truth of the matter but to determine whether there is a genuine issue for trial." Id. at 249, 106 S. Ct. 2505.

III

A

California trade secret law governs UniRAM's misappropriation claim. See Doc. # 375 at 4. Under the California statute, trade secret misappropriation is defined, in part, as:

(2) Disclosure or use of a trade secret of another without express or implied consent by a person who:

* * *

(B) At the time of disclosure or use, knew or had reason to know that his or her knowledge of the trade secret was:

* * *

(ii) Acquired under circumstances giving rise to a duty to maintain its secrecy or limit its use; or

(iii) Derived from or through a person who owed a duty to the person seeking relief to maintain its secrecy or limit its use; * * *.

Cal. Civ. Code § 3426.1(b). UniRAM claims its trade secrets consisted of particular combinations of features and that UniRAM communicated these secrets to TSMC through tape outs Shau shared with TSMC who then disclosed these secrets to MoSys. TSMC denies that the secrets were ever communicated to TSMC and that the court can make this determination as a matter of law.

TSMC primarily argues that UniRAM's disclosures to TSMC consist of tape outs which merely list dozens of different possible circuit "features" and do not specify the unique combinations of features that make up the trade secrets at issue. According to TSMC, summary judgment is appropriate because "not a single document UniRAM relies on to support disclosure of its alleged trade secrets recites all of the features for a given combination secret." Doc. # 375 at 7. Moreover, according to TSMC, none of Shau's oral disclosures to TSMC ever identified specific secret combinations. Id. at 9. TSMC's theory is that UniRAM must disclose its elements and secret combinations in "a single form or on a single occasion." Id. at 10; compare RM Cummings, Some Aspects of Trade Secrets and Their Protection: The Public Domain and the ``Unified Description' Requirement, 54 Ky. L. J. 190, 191 (1966).

In essence, TSMC argues that when an alleged trade secret consists of a combination of nonsecret elements, plaintiffs have a claim only if they can prove they disclosed to defendants the precise combination that constitutes the secret. TSMC provides no authority for this proposition. The issue is one of disclosure, and the authorities TSMC cites are not relevant in determining the level of legally required disclosure. The rulings in Julie Research Labs., Inc. v. Select Photographic Eng'g, Inc., 810 F. Supp. 513, 519-20 (S.D.N.Y. 1992), Cybertek Computer Prods., Inc. v. Whitfield, 203 U.S.P.Q. 1020, 1024 (Cal.App.Super.1977), IDX Sys. Corp. v. Epic Sys. Corp., 285 F.3d 581 (7th Cir. 2002), and Am. Airlines, Inc. v. KLM Royal Dutch Airlines, Inc., 114 F.3d 108 (8th Cir.1997) do not state as broad a principle as TSMC contends.

Julie Research and Cybertek offer only the uncontroversial proposition that a combination of nonsecret elements is entitled to the same trade secret protection as any discovery. In IDX, the court rejected the plaintiff's extremely vague description of its trade secrets, but the trade secrets at issue here have been described adequately. And American Airlines was an estoppel case in which the court held that once the plaintiff contended that his trade secret was a combination of five elements, he could not change course in the middle of litigation and, in a "sham" attempt to avoid summary judgment, assert that his secret involved a combination of only four elements. See American Airlines, 114 F.3d at 111-12. The level of requisite disclosure was never at issue in those cases. Accordingly it is not true as a matter of law that UniRAM's claim will survive only if it disclosed to TSMC the exact combinations that make up its trade secrets.

TSMC's failure to locate any authority in support of its position may not be entirely its fault. It appears the question of knowledge of a combination trade secret may be a case of first impression. One pair of recent commentators "found no published decisions analyzing the knowledge element of a plaintiff's combination trade secret claim." Tait Graves & Alexander Macgillivray, Combination Trade Secrets and the Logic of Intellectual Property, 20 Santa Clara Computer & High Tech. L. J. 261, 285-86 & n. 63 (2004).

Nevertheless, courts have addressed a similar question in a related context, and that analysis is persuasive here. Courts must often determine whether a combination trade secret is a secret when the elements of the combination have been disclosed publicly but the combination itself has not. In that context, courts have held that knowledge of the combination may be presumed as long as one skilled in the art could view the nonsecret elements and replicate the combination without undue difficulty. See Computer Care v. Serv. Sys. Enters., Inc., 982 F.2d 1063, 1073 (7th Cir.1992); Pope v. Alberto-Culver Co., 296 Ill.App.3d 512, 230 Ill. Dec. 646, 694 N.E.2d 615, 618-19 (1998); Ashland Management Inc. v. Janien, 82 N.Y.2d 395, 407-08, 604 N.Y.S.2d 912, 624 N.E.2d 1007 (N.Y.1993). This principle means that in the present case, where the elements have only been disclosed to another party in a confidential relationship, the defendant might be deemed to have constructive knowledge of a combination of elements even though only the separate elements have been disclosed.

To illustrate the analogy, in Servo Corp. of America v. General Elec. Co., 393 F.2d 551 (4th Cir.1968), the Fourth Circuit decided the question whether a trade secret loses its secret status if its elements have been disclosed publicly or only if the combination has been disclosed publicly. Servo had claimed General Electric misappropriated its design for railroad "hot box detectors." Id. at 552. Servo argued that even though each of the elements of its design was publicly disclosed, the secret combination had not yet been made public because the components of the combination had not been disclosed within "a single integrated document." Id. at 554. The court rejected that contention, which it likened to the doctrine of anticipation in patent law. Id. Instead, the question was "whether, taking into account all of the plaintiff's relevant disclosures, it is reasonable to conclude that a competitor could have ascertained the working combination from an examination of those disclosures." Id. The court held in the affirmative, essentially ruling that the general public would be charged with knowledge of the combination. See id. The same principle applies here—knowledge of the combination may be imputed to those with knowledge of the components. If TSMC had knowledge of the individual features comprising Shau's invention, then knowledge of the combination may be imputed as long as it is reasonable. Insisting on a unified description in a single integrated document, as TSMC asserts, is unnecessary. Accordingly, even if TSMC is correct that the tape outs and discussions disclose only elements and not combinations, TSMC might have knowledge of UniRAM's trade secrets if inferring such knowledge would be reasonable under the circumstances.

TSMC's position does not lack appeal. Simple and clear rules of law are always appealing, especially to nonlawyers. See very generally Dennis Jacobs, The Secret Life of Judges, 75 Fordham L. Rev. 2855 (2007). Here, however, the problem should be categorized as a question of fact. The issue turns on how easy or difficult it is to assemble the relevant elements into the secret combination. Many factors will go into that determination, and those factors should be weighed by a jury. As an example, if the Coca-Cola formula consists of a combination of ten nonsecret ingredients, then a document which lists those ten ingredients in the middle of a sea of other chemicals might not constitute adequate disclosure of the formula because the precise recipe would be too difficult to derive. In that instance, it may be unfair to hold accountable trade secret defendants for picking out the ten ingredients from a very long list and contending that those ten and only those ten would have some special benefit once combined. On the other hand, if the list has only eleven ingredients, and if one skilled in the relevant art or science viewing the list would likely know that the formula was a combination of ten of the ingredients, then the matter is quite different. See Graves & Macgillivray, supra, at 283 (stating that a "combination [might be] obvious because of the limited number of potential alternatives"). The same may be true if the list contains all the correct ingredients but does not disclose the precise amount of each chemical that must be added to the mix. There is no threshold of specificity that must be met before the trade secret will be deemed "disclosed" as a matter of law. In short, a defendant's "knowledge" of a secret combination will depend on how easy or difficult it is to piece together. That is a question of fact.

The problem is not purely an issue of permutations and combinations. Qualitative factors are just as important as probabilistic ones. Certain factors or features may be naturally related to one another, may be completely incompatible, may be standard in the industry, may be obvious to those skilled in the art or might otherwise simplify (or complicate) the problem of discovering the secret combination. The DRAM invention at issue here illustrates the concept because certain features imply others. See, for example, Doc. # 442 Ex. I (Mandelman Decl.) at 13 ("Embedding DRAM in logic naturally calls for a hidden refresh and a SRAM interface * * *."). Just as some combinations are so widespread as to no longer be secrets, those same combinations might whittle down the possibilities and thereby reduce the difficulty of picking out the truly secret combinations. TSMC's method of simply counting up the potential combinations will not do. The jury needs to weigh these fact-intensive circumstances.

The above discussion presumes that UniRAM's disclosures disclose only elements and not specific secret combinations, but of course UniRAM might still be able to demonstrate at trial that the tape outs do in fact disclose specific combinations. In any event, TSMC's knowledge of the alleged trade secrets is a question of fact to be decided at trial.

B

TSMC next contends that UniRAM has offered no evidence that "TSMC took UniRAM's alleged 12 trade secret combinations and passed them to MoSys" (see Doc. # 409 (TSMC reply) at 3) or to Matsushita (see id. at 5). Plaintiffs alleging trade secret misappropriation may prove such misappropriation by circumstantial as well as direct evidence. See Droeger v. Welsh Sporting Goods Corp., 541 F.2d 790, 792 (9th Cir.1976); Pioneer Hi-Bred Intern. v. Holden Foundation Seeds, Inc., 35 F.3d 1226 (8th Cir.1994); Comprehensive Technologies Intern., Inc. v. Software Artisans, Inc., 3 F.3d 730, 736 (4th Cir.1993). Circumstantial evidence is particularly appropriate in trade secret cases:

It is well recognized with respect to trade secrets that:

[m]isappropriation and misuse can rarely be proved by convincing direct evidence. In most cases plaintiffs must construct a web of perhaps ambiguous circumstantial evidence from which the trier of fact may draw inferences which convince him that it is more probable than not that what plaintiffs allege happened did in fact take place. Against this often delicate construction of circumstantial evidence there frequently must be balanced defendants and defendants' witnesses who directly deny everything.

Q-Co Industries, Inc. v. Hoffman, 625 F. Supp. 608, 618 (S.D.N.Y.1985).

UniRAM has offered circumstantial evidence sufficient to create a genuine issue of fact for trial. According to UniRAM, MoSys was a failed player in the DRAM market and by 1998 owed tens of millions of dollars to TSMC, one of its shareholders. See Doc. # 442 Ex. B at 514, 518; id. Ex. C at 088; id. Ex. D at 570-71. Then in 1998, top officials at both companies met to find a solution. See id. Ex. E at 078-81. Shortly thereafter (and after TSMC became privy to Shau's invention), TSMC "was working jointly with MoSys to develop an embedded logic process that mirrored Dr. Shau's technology." Doc. # 395 (UniRAM Opp.) at 5; Doc. # 442 Ex. F at MOS168157-58; see also Doc. # 395 at 14-15 (detailing changes to MoSys products after its meeting with TSMC). These facts, particularly the 1998 meeting, go beyond the factual allegations in the complaint. See Doc. # 16 at ¶¶ 32-38. This narrative is adequate to support UniRAM's theory of misappropriation. Whether or not TSMC actually disclosed Shau's secrets to MoSys at the 1998 meeting is a question for the jury.

The same cannot be said, however, of any claim that TSMC disclosed trade secrets to Matsushita. UniRAM has offered no evidence of any communications between TSMC and Matsushita. Doc. # 375 at 10. Accordingly there is no basis for a reasonable inference that TSMC improperly disclosed UniRAM's secrets to Matsushita.

C

Apart from its claim that UniRAM never disclosed any trade secrets, TSMC claims that four specific TSMC products do not utilize the special "small block architecture" feature claimed by UniRAM as an element in its secret combinations. Accordingly, TSMC argues that it did not misappropriate any of UniRAM's trade secrets which include that feature—secrets numbers 1, 2, 4, 5, 6 and 7. Doc. # 375 at 12.

Specifically, TSMC argues that UniRAM's definition of the term "small block architecture" on its face does not apply to TSMC parts numbers TMF167, TMB138, TMF964 or TM9821. Doc. # 375 at 11. In its definitions of terms, UniRAM describes small block architecture as:

A memory architecture in which millions of DRAM memory cells are divided into hundreds or thousands of independent blocks whose word lines and bit lines are shorter than (and exhibit less capacitance than) those in conventional DRAM architectures.

See Doc. # 444, Tuttle Decl. Ex. Q (emphasis added). The parties do not appear to dispute the following characteristics of four TSMC products:

(1) TMF167 has 1.081 million cells and 66 blocks;

(2) TMB138 has 1.081 million cells and 32 blocks;

(3) TMF964 has 132 blocks; and

(4) TM9821 has 132 blocks.

Doc. # 375 at 14-16. From these data, TSMC argues that the first two products do not use small block architecture because 1.081 million cells does not rise to the level of "millions" (plural) of cells. Id. Moreover, TSMC argues that 66 blocks and 32 blocks do not rise to the level of "hundreds or thousands of blocks." Id. Along those lines, TSMC asserts that the last two products have only 132 blocks and therefore do not contain "hundreds" (plural) of blocks. Id. at 16 ("One hundred and thirty-two plainly does not fall within the range of hundreds to thousands. Just as one dollar and thirty-two cents is not dollars. Simply put, 132 is not hundreds."). TSMC's position is that the word hundreds, "as the plural of hundred, * * * refers to quantities of 200 or more." Id. at 17. Presumably the same construction applies to millions. UniRAM responds by arguing that "hundreds or thousands" was intended as an illustration of the relevant order of magnitude rather than a formal limitation. See Doc. # 395 at 23.

One might theorize endlessly upon TSMC's grammatical interpretations. One dollar and thirty-two cents might be restated into the plural form as "one point three two dollars." The Oxford English Dictionary defines "plural" as "denoting more than one," which supports UniRAM because even "one and one half" hundreds is more than a single hundred. UniRAM's expert takes a similar position (see Doc. # 375 at 17).

In any event, the court declines to base its decision on the number of angels which might fit upon a pinhead (or upon a DRAM circuit for that matter). Instead, the court rejects TSMC's contentions because any asserted deviations from the specified definition are not material. See Speech Tech. Assoc. v. Adaptive Comm. Sys., Inc., No. C-88-2392-VRW, 1994 WL 449032 at *9-10 (N.D.Cal., Aug. 16, 1994) ("The incidental differences between Prototype # 1 and the redesigned Alltalks do not absolve defendants from liability for misappropriation of trade secrets."); see also Am. Can Co. v. Mansukhani, 742 F.2d 314, 328-29 (7th Cir.1984) ("If the law were not flexible enough to reach [independent] modifications, trade secret protection would be quite hollow"). The issue whether the four TSMC products are "substantially derived" (see id.) from UniRAM's trade secrets is a question of fact for precisely the same reasons that misappropriation is a question of fact, as discussed in part IV. Drawing all reasonable inferences in UniRAM's favor, the court declines to hold as a matter of law that TSMC did not misappropriate any combinations including small block architecture.

IV

TSMC asserts that UniRAM's trade secrets claims are barred by California's three-year statute of limitations (see Cal. Civ. Code § 3426.6) and that the remaining unfair competition and breach of contract claims are barred by the four-year statute of limitations (see Cal. Bus & Prof. Code § 17208; Cal. Code Civ. Proc. § 337). UniRAM responds that it was not until April 25, 2001 at the earliest (two years and fifty-one weeks before the cutoff) that the statute of limitations began to run. See Doc. # 391 at 24.

California is an inquiry notice state when it comes to the statute of limitations. The limitations period begins to run when the plaintiff knows of the injury or should know of the injury. See Grisham v. Philip Morris USA, Inc., 40 Cal. 4th 623, 634, 54 Cal. Rptr. 3d 735, 151 P.3d 1151 (2007); Fox v. Ethicon Endo-Surgery, Inc., 35 Cal. 4th 797, 807, 27 Cal. Rptr. 3d 661, 110 P.3d 914 (2005). Plaintiffs are deemed to have this constructive knowledge "only where there is a duty to inquire, as where plaintiff is aware of facts which would make a reasonably prudent person suspicious." Hobart v. Hobart Estate Co., 26 Cal. 2d 412, 438-39, 159 P.2d 958 (1945). California has expanded upon the "reasonable suspicion" standard by stating that the plaintiff must suspect "that someone has done something wrong to him, ``wrong' being used, not in any technical sense, but rather in accordance with its lay understanding. * * * He has reason to suspect when he has notice or information of circumstances to put a reasonable person on inquiry * * *." Norgart v. Upjohn Co., 21 Cal. 4th 383, 397-98, 87 Cal. Rptr. 2d 453, 981 P.2d 79 (1999) (citing Jolly v. Eli Lilly & Co., 44 Cal. 3d 1103, 1110, 245 Cal. Rptr. 658, 751 P.2d 923 (Cal.1988)). The plaintiff must have some basis for making further inquiries—in practice this may mean that if a plaintiff knows he has been injured but does not know who has injured him, he has a duty to inquire, which starts the statute of limitations. See, for example, McKelvey v. Boeing North American, Inc., 74 Cal. App. 4th 151, 160-61, 86 Cal. Rptr. 2d 645 (1999) (barring plaintiff's claim under the statute of limitations because the plaintiff had been injured by contamination and should have known of media reports linking the contamination to Boeing); Jolly, 44 Cal.3d at 1113, 245 Cal. Rptr. 658, 751 P.2d 923 (stating that the plaintiff believed early on that DES was a defective drug, thus triggering a duty to inquire or file a Doe lawsuit). The party with a duty to inquire is charged with the results of what a reasonable investigation would uncover. Accordingly, UniRAM's claims against TSMC will be time-barred if it had information to suspect that it had been wronged.

TSMC points to four documents dated outside the period which it claims indisputably triggered UniRAM's duty to inquire. Because, however, each of those documents presents only a factual question whether UniRAM should have inquired further, summary judgment is inappropriate.

First, TSMC argues that an email sent on November 20, 2000, from TSMC employees to Shau and Sidney Yen at UniRAM establishes that they knew TSMC was producing DRAM chips with logic processes. The email stated that TSMC would no longer be utilizing the Emb-DRAM process for producing chips, and it also stated that the sender of the email, Snitsky, would be "responsible for 0.13 1T SRAM." Doc. # 376 at 13-14. TSMC contends that the email could only mean that TSMC would begin building 1T-SRAM devices in the 0.13um logic process, in which case TSMC would be using UniRAM's trade secrets. Id. at 14. UniRAM disputes that the email could only mean that TSMC would be using a logic process. UniRAM asserts that TSMC had been utilizing many methods of building DRAM devices, and therefore the cancellation of one method (EmbDRAM) would not necessarily imply that the 1T-SRAM devices would be built using a standard logic process. Doc. # 391 at 14. Accordingly, the parties dispute whether the November email would have communicated to UniRAM that TSMC had misappropriated trade secrets in producing its own version of Shau's invention.

Second, TSMC claims that a January 2001 email proves Shau knew that TSMC's product—1T SRAM—was the misappropriated version of his own product, DRAM. Doc. # 376 at 14-15. TSMC claims Shau spoke in an email of "1-T SRAM, which is actually DRAM." According to TSMC, 1T-SRAM utilized many of the secret features of DRAM, and therefore Shau's email was his implicit recognition that 1T-SRAM was the misappropriated version of DRAM. Id. at 15. UniRAM responds that the term "1T-SRAM" is only a generic term that refers to DRAM generally. Shau decl. at ¶ 27. Shau claims that the "1T" in "1T-SRAM" means only that the device is manufactured using one transistor. Id. One feature of DRAM is that it is produced using single transistor memory cells. Id. Accordingly, Shau understood 1T-SRAM "as a more general term by TSMC engineers to describe embedded DRAM." Id. The parties are in clear disagreement whether the January email demonstrates that UniRAM knew TSMC may have misappropriated its secrets or whether it only shows two phrases which describe the same thing.

Third, TSMC claims MoSys sent UniRAM a threat letter in February 2001 in which MoSys accused UniRAM of infringing MoSys patents covering logic-embedded DRAM. Doc. # 376 at 15-16. Based on that letter, TSMC argues, UniRAM was aware that MoSys was using several of UniRAM's accused features. Id. at 15. The letter attached the relevant patent, which was titled "Memory Cell For DRAM Embedded In Logic." Id. TSMC asserts that because the patent contained the phrase "embedded in logic" in the title, UniRAM knew that the MoSys product was manufactured in a logic process (which would constitute misappropriation). UniRAM responds that the title of the patent was not relevant because the patent claims "had nothing to do with any of the UniRAM trade secret features." Shau decl. at ¶ 29. Accordingly UniRAM had no reason to be suspicious of any misappropriation. Id. Shau claims he did not suspect TSMC of any wrongdoing: "[T]he mere fact that another company [MoSys] was attempting to employ a logic process for embedded DRAM would not have caused me to suspect misappropriation, especially given that the patent was issued to MoSys rather than TSMC, the party to whom I had disclosed trade secrets." Id. Shau was free to believe that MoSys was developing a similar product independently and that he was not the victim of any wrongful conduct. He did not necessarily have information that MoSys appropriated his trade secrets through TSMC or anyone else. Shau's credibility on this point is for the jury to determine. Overall, the parties genuinely dispute whether the February letter shows that UniRAM knew TSMC may have misappropriated its trade secrets.

Lastly, TSMC claims that UniRAM employees attended a conference hosted by TSMC in 2000 during which TSMC disclosed that MoSys, one of its customers, was using UniRAM's trade secrets. UniRAM's second-in-command, Sidney Yen, attended the conference. The conference included a slide show presentation with the following slide in the "Embedded DRAM" portion: "MoSys Corp.: 16Mb, 1T-SRAM architecture, SSRAM Interface, granularity 128Kb, 200MHz." Doc. # 375 at 17. TSMC contends that the "presentation made clear" to Yen (and therefore to UniRAM) "that TSMC and MoSys were collaborating on 1T-SRAM, which had an SRAM Interface, like UniRAM's alleged trade secret technology." Id. Shau disputes the "clarity" of the presentation materials, noting that the "entire embedded DRAM portion of [the symposium] relates to EmbDRAM" and not to embedded DRAM made using a logic process as would be relevant to his trade secrets. Shau decl. at ¶ 31, 30. Moreover, the slide specifies a granularity of 128Kb, which may suggest a type of architecture that Shau rejected. Shau claimed the slide "would lead [him] into thinking they were still using old DRAM." Id. at ¶ 31. The clarity of the presentation, and what UniRAM understood from it, is a factual question. Resolution of this dispute on a motion for summary judgment is improper.

Because there is a genuine issue of fact as to when UniRAM knew it had been injured or wronged, summary judgment on statute of limitations grounds is improper.

V

For the foregoing reasons, the court GRANTS TSMC's motion for summary judgment as to Matsushita products and DENIES the motion in all other respects. Because the disputed portions of the Shau and Murphy declarations are not essential to the court's holding, TSMC's objections to the statements therein (see Doc. # 410) are moot.

IT IS SO ORDERED.