Comet Technologies USA, Inc. v. Xp Power, LLC

FOR PUBLICATION UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT COMET TECHNOLOGIES USA, Nos. 23-15601, INC., a Delaware corporation; 23-15709, 25-745 COMET, AG, a Swiss corporation; D.C. No. YXLON INTERNATIONAL, 5:20-cv-06408- GMBH, a German corporation, NC Plaintiffs – Appellees/ Cross-Appellants, v. OPINION XP POWER, LLC, a California Limited Liability Company, Defendant – Appellant/ Cross-Appellee. Appeals from the United States District Court for the Northern District of California Nathanael M. Cousins, Magistrate Judge, Presiding Argued and Submitted September 19, 2025 San Francisco, California Filed July 14, 2026 2 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC Before: David F. Hamilton, Ryan D. Nelson, and Patrick J. Bumatay, Circuit Judges. * Opinion by Judge Hamilton; Concurrence by Judge Hamilton; Dissent by Judge Bumatay SUMMARY ** Trade Secret Misappropriation The panel (1) reversed the district court’s judgment after a jury trial in favor Comet Technologies USA, Inc., and affiliates on their claims against XP Power, LLC, for misappropriating trade secrets and (2) remanded for a new trial. Comet alleged that XP, a fellow manufacturer of components used in fabricating computer chips, misappropriated Comet’s trade secrets in violation of the Defend Trade Secrets Act. The jury awarded Comet $40 million in compensatory and punitive damages, a permanent injunction against any further use of the alleged trade secrets by XP, and over $17 million in attorney fees. The panel held that the district court erred by instructing the jury that XP bore the burden of disproving an essential * The Honorable David F. Hamilton, United States Circuit Judge for the Court of Appeals, Seventh Circuit, sitting by designation. ** This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader.  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 3 element of Comet’s DTSA claims, namely, that its trade secrets were not readily ascertainable by proper means. The panel concluded that XP did not invite the error because it made a timely and correct objection on the burden of proof. The panel further concluded that the error was not harmless because the jury instructions were not accurate as a whole, and conflicting evidence on many issues of ready ascertainability and the ensuing damage calculations presented the sorts of disputes that a properly instructed jury must decide. The panel addressed an evidentiary issue in a separate memorandum disposition. Concurring, Judge Hamilton wrote that he would address a challenge to the relief awarded and would join the Third and Seventh Circuits in holding that awarding both damages for unjust enrichment based on avoided costs and injunctive relief under the DTSA does not create an impermissible double recovery. Dissenting, Judge Bumatay agreed with the majority that the district court used an erroneous jury instruction, but he would hold that the error was harmless because the wealth of undisputed evidence introduced at trial easily established that it was more probable than not that the jury would have reached the same verdict had it been properly instructed. 4 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC COUNSEL Jason M. Wilcox (argued), Joseph M. Capobianco, and John C. O’Quinn, Kirkland & Ellis LLP, Washington, D.C.; Adam R. Alper, Kirkland & Ellis LLP, San Francisco, California; Michael W. De Vries and Sharre Lotfollahi, Kirkland & Ellis LLP, Los Angeles, California; Elizabeth Nielson, Kirkland & Ellis LLP, Salt Lake City, Utah; Steven J. Lindsay, Kirkland & Ellis LLP, Chicago, Illinois; for Plaintiffs-Appellees. Michael E. Bern (argued), Uriel Hinberg, and Christine C. Smith, Latham & Watkins LLP, Washington, D.C.; Patricia Young, Latham & Watkins LLP, Menlo Park, California; for Defendant-Appellant. Robert J. Gunther Jr., Wilmer Cutler Pickering Hale and Dorr LLP, New York, New York, for Intervenor.  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 5 OPINION HAMILTON, Circuit Judge: These appeals arise from a dispute between two manufacturers of components used in fabricating computer chips. Comet Technologies USA, Inc. and affiliates sued XP Power, LLC under federal and state law for misappropriating trade secrets. Comet won a $40 million verdict for compensatory and punitive damages, a permanent injunction against any further use of the alleged trade secrets by XP, and over $17 million in attorney fees. We reverse the judgment and remand for a new trial. The district court erroneously instructed the jury that XP bore the burden of disproving an essential element of Comet’s claims, namely, that its trade secrets were not readily ascertainable by proper means. Because we vacate the judgment below, we do not reach XP’s challenges to the awarded relief in this opinion. 1 I. Factual and Procedural Background A. The Theft Christopher Mason and two other senior engineers left their jobs at Comet in February 2018 to work for XP on developing new product lines in the radio frequency power generator and impedance matching market. They brought with them not only years of industry experience but also 1 In a separate non-precedential memorandum, we also reject XP’s argument that the district court abused its discretion in excluding certain evidence under Federal Rule of Evidence 403. Judge Bumatay dissents from this opinion’s order of a new trial based on the jury instruction issue. In a separate concurring opinion, Judge Hamilton addresses XP’s challenges to the relief awarded. 6 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC thousands of confidential documents and files detailing Comet’s products, research and development strategy, and the underlying technologies. Stolen documents in hand, they delivered complete designs and a development plan for XP’s new product lines in just nine days. Comet quickly discovered the theft and filed suit, alleging violations of both the federal Defend Trade Secrets Act (DTSA), 18 U.S.C. § 1836 et seq., and the California Uniform Trade Secrets Act (UTSA), Cal. Civ. Code § 3426 et seq. The DTSA, enacted in 2016, applies to misappropriation of trade secrets used in interstate or foreign commerce. It provides federal remedies that supplement, but do not replace, state trade secret laws. See 18 U.S.C. § 1838; Quintara Biosciences, Inc. v. Ruifeng Biztech, Inc., 149 F.4th 1081, 1088 (9th Cir. 2025); Attia v. Google LLC, 983 F.3d 420, 424–25 (9th Cir. 2020). The parties and the district court narrowed the issues before and during trial. Comet originally alleged misappropriation of twenty alleged trade secrets. At the district court’s sensible urging, Comet agreed to pursue only five at trial, designated as follows: • Trade Secret D: Da Vinci RF Generator Control, Digital Measurement, and Software. • Trade Secret E: Next Generation RF Matching Network. • Trade Secret L: Kiyo Matching Network. • Trade Secret S: AMAT Matching Network.  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 7 • Trade Secret T: Comet’s Manufacturing, Sales, and Pricing. 2 During trial, Comet also voluntarily dismissed its California UTSA claims to streamline the case, leaving only those brought under the federal DTSA. That was a significant change that led to the reversible instructional error that we explain in Section III. B. The Trial The jury heard from numerous fact witnesses, including a video deposition of Mason, and technical and damages experts from both parties. The evidence included one proverbial smoking gun: a call between Mason and a headhunter for XP, a call in which Mason—still employed at Comet—offered a “crew” to provide a “turnkey” (meaning “already done”) product design. In what one Comet expert, Dr. Stanley Shanfield, aptly called a “remarkable” move, the headhunter recorded the call and immediately forwarded it to XP’s chief executive officer and 2 In Comet’s words: These valuable RF technologies are used in the semiconductor industry to manufacture silicon computer chips, which are used around the world. An RF generator generates radio frequency power, and an RF matching network (sometimes called a “match” or “matchbox”) receives the generated RF power, and conditions it so it can be used to control plasma, which is in turn used to carve a silicon wafer into a semiconductor. For an introduction to semiconductor manufacturing, see Corey Richard, Understanding Semiconductors: A Technical Guide for Non-Technical People 57–80 (2023). 8 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC other senior executives, telling them “please keep this recording strictly confidential. You will see why.” On damages, Comet asked the jury to award the full research and development costs that XP avoided having to spend by stealing the alleged trade secrets instead of developing them independently. Comet relied on this theory of unjust enrichment, rather than seeking its own lost profits or XP’s ill-gotten gains. By the time of trial, XP had not yet released any new products using the Comet secrets. The district court instructed the jury that Comet had the burden of showing that: (1) it owned the alleged trade secrets, (2) XP misappropriated them, and (3) they were in fact trade secrets at the time of the misappropriation. Over XP’s objection, however, the district court also instructed the jury that XP bore the burden of proving its so-called “affirmative defense” that the alleged trade secrets were “readily ascertainable by proper means.” C. The Verdict and Judgment In a special verdict, the jury found that Trade Secrets D, E, L, and S were owned by Comet and protected as trade secrets under the DTSA, and that XP had misappropriated Trade Secrets D, E, and L, but not Trade Secret S. The district court had already granted XP judgment as a matter of law on Trade Secret T. 3 The jury awarded $5 million in damages for misappropriation of Trade Secret D and $15 million for misappropriation of Trade Secret E, but nothing for Trade Secret L, for a total of $20 million in compensatory damages. 3 The district court wrote regarding Trade Secret S that the jury “appears to have credited XP’s argument that . . . the information was available in non-secret forums or had previously been disclosed outside Comet.”  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 9 It also found that XP “willfully and maliciously” misappropriated at least one of the trade secrets and awarded a further $20 million in punitive damages for a total award of $40 million. After trial, the district court granted Comet’s motion for a permanent injunction requiring XP to: refrain from using, disclosing, selling, or distributing Trade Secrets D, E, and L; remove and quarantine the stolen information; and allow third-party audits to ensure compliance. Later, the district court awarded Comet over $17 million in attorney fees. XP appealed, raising several issues with the trial and judgment. Comet conditionally cross-appealed regarding only the award of zero damages for Trade Secret L. We consolidated the cross-appeals on the merits and XP’s later appeal of the fee award. II. Appellate Jurisdiction The district court had subject matter jurisdiction under 28 U.S.C. § 1331 and 18 U.S.C. § 1836(c). We have appellate jurisdiction under 28 U.S.C. § 1291. XP’s opening brief on appeal questioned our appellate jurisdiction on the basis that, when XP filed its original notice of appeal, there was no final judgment because the district court had not yet ruled on Comet’s motion for prejudgment interest. Any potential jurisdictional defect was cured when XP filed an amended notice of appeal after the district court decided the motion for prejudgment interest in January 2025. See Fed. R. App. P. 4(a)(4)(B)(i); In re Jack Raley Construction, Inc., 17 F.3d 291, 294 (9th Cir. 1994) (“[T]he prudent course of action is merely to file a fresh appeal after entry of final judgment.”); see also Martinez v. Barr, 941 F.3d 907, 916 (9th Cir. 2019) (“We 10 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC prioritize substantive rights of parties over procedural defects in appeals, allowing premature appeals to ripen absent any prejudice to the appellee.”). The parties correctly agreed that this course would establish jurisdiction, and neither argues prejudice from the fact that some briefing had been filed before then. We have jurisdiction and proceed to the merits. III. The Erroneous Jury Instruction Jury instructions in civil trials focus the jury on the questions they must decide and how they should go about deciding them. Trial judges enjoy wide latitude in crafting jury instructions, but they must get the substance correct. Here, we must reverse and remand for a new trial. We agree with both plaintiffs and defendants that the district court improperly instructed the jury on the burden of proof for lack of ready ascertainability, an essential element of Comet’s DTSA claims. XP did not invite the error, and Comet has not convinced us that the error was harmless. A. How It Happened Instruction 20 told the jury that XP was not liable for misappropriation if XP proved by a preponderance of the evidence that the alleged trade secrets were readily ascertainable by proper means, meaning they could be lawfully “obtained, discovered, developed, reverse- engineered, or compiled without significant difficulty, effort, or expense.” This instruction was erroneous because the DTSA instead puts the burden of proof on the plaintiff to show that its alleged trade secrets were not readily ascertainable by proper means. See 18 U.S.C. § 1839(3)(B). Instruction 20 thus erroneously flipped the burden of proof on an essential element of Comet’s claim.  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 11 Some states, including California, put the burden of proof on the defendant by making ready ascertainability an affirmative defense to a trade secret claim. California Civil Jury Instructions No. 4420 (2025). Indeed, to avoid confusing the jury, XP had originally proposed Instruction 20 at a time when Comet was still asserting claims under both the federal DTSA and California’s UTSA. Several days into trial, however, Comet voluntarily dropped its California UTSA claims to streamline the issues for the jury. The district court asked the parties to explain how this change would affect the jury instructions and verdict form. In response, XP said correctly that Instruction 20 should be amended to shift the burden of proof on lack of ready ascertainability to Comet, in line with the federal DTSA. Comet objected and argued that Instruction 20 should be removed entirely, claiming quite erroneously that the “DTSA does not permit a defendant to avoid liability for misappropriation by showing that it could have found the information some other way.” The district court rejected both requests without an on-the-record explanation. At the final charging conference the next day, the district court asked if there were “any other suggestions from either side as to the verdict form as to the issues not previously objected to in your writings?” Comet asked the district court to remove the “not readily ascertainable” question from the verdict form, repeating in substance its written objection to Instruction 20. XP responded that “I believe we’ve raised all our concerns already and they’re preserved,” plainly referring to the previous day’s motion and denial. Then, responding to Comet’s request, XP argued that Instruction 20 and the corresponding verdict question were “appropriate under the DTSA.” The district court agreed and left them in. 12 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC We review a district court’s formulation of civil jury instructions for abuse of discretion, but we review de novo whether instructions substantively misstate the law. Chess v. Dovey, 790 F.3d 961, 970 (9th Cir. 2015). The parties agree that Instruction 20 erroneously reversed the burden of proof. Comet raises two arguments to save the verdict from the error: (1) that XP invited the error, and (2) that the error was harmless. We reject both arguments. B. No Invited Error Comet argues that XP invited the instructional error when it told the district court in the final charging conference that Instruction 20 should be given. A party waives appellate review by inviting an error when it is aware of the applicable law but nonetheless “intentionally relinquishe[s] or abandon[s] a known right” by advancing the error or failing to object. United States v. Perez, 116 F.3d 840, 845 (9th Cir. 1997) (en banc). Proposing a flawed jury instruction is a “paradigmatic example” of invited error. United States v. Magdaleno, 43 F.4th 1215, 1220 (9th Cir. 2022). The requisite intent to invite an error can be inferred when a party insists on a faulty instruction despite the court or the other side pointing out the problem. See United States v. Guthrie, 931 F.2d 564, 567 (9th Cir. 1991) (“Here, the district court offered to instruct the jury on what it could infer . . . . Guthrie declined the offer, however, and proposed instead that the court simply reread its more general instruction on the element of knowledge. The court complied with Guthrie’s request, so he cannot now complain of what he received.”); United States v. Baldwin, 987 F.2d 1432, 1437 (9th Cir. 1993) (finding invited error where criminal defendant opposed government’s request to give the erroneously missing jury instruction); see also Fed. R. Civ.  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 13 P. 51(d)(1)(A) (“A party may assign as error . . . an error in an instruction actually given, if that party properly objected . . . .”). XP did not invite this error. Far from acquiescing, XP objected immediately when Comet dismissed its state law claims and thereby made Instruction 20 erroneous. XP specifically drew the district court’s attention to the different burdens of proof. In response, Comet muddled the issue by asking the district court to remove the instruction entirely on the erroneous theory that XP did not claim to have actually reverse-engineered the information, an argument that Comet has wisely abandoned. The district court’s failure to correct the error as XP requested was not XP’s fault. 4 XP’s statements at the charging conference the next day also do not show invited error. The district court asked the parties for objections to the verdict form that were not previously raised in writing; XP’s objection to Instruction 20 was neither. Even so, XP did not remain silent. It reminded the judge of its previous objection and noted that the issue 4 Ninth Circuit precedent and at least one California state court case support Comet’s argument as to the California UTSA claims. See California Civil Jury Instructions No. 4420 (2025), citing ABBA Rubber Co. v. Seaquist, 235 Cal. App. 3d 1, 21 n.9 (1991). The same is not true of the DTSA, which defines a “trade secret” to exclude information that is not “readily ascertainable through proper means.” 18 U.S.C. § 1839(3)(B); e.g., Masimo Corp. v. True Wearables Inc., No. 18-cv- 2001, 2021 WL 2548690, at *3 (C.D. Cal. Apr. 28, 2021) (“[T]here is one key difference between the [California UTSA] and the DTSA . . . . [U]nder the [California UTSA], ready ascertainability is only a defense insofar as the defendant actually gained knowledge of the trade secret by use of those materials which make the trade secret readily ascertainable.”), aff’d, No. 21-2146, 2022 WL 205485 (Fed. Cir. Jan. 24, 2022), citing IMAX Corp. v. Cinema Technologies, Inc., 152 F.3d 1161, 1168 n.10 (9th Cir. 1998). 14 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC was preserved. More fundamental, the district court’s erroneous ruling the day before and Comet’s proposal to remove Instruction 20 entirely forced XP into a Hobson’s choice between either a flawed instruction on ready ascertainability or no instruction at all. Having made a timely and correct objection on the burden of proof the day before, XP’s argument in the alternative—to keep an imperfect instruction on an element important to XP’s defense—did not amount to invited error. See United States v. Laurienti, 611 F.3d 530, 544 (9th Cir. 2010) (“[N]othing prevents a party from arguing in the alternative.”). 5 C. No Harmless Error An instructional error in a civil case does not require reversal if the party defending the judgment shows that the error was “more probably than not harmless.” Clem v. Lomeli, 566 F.3d 1177, 1182 (9th Cir. 2009), quoting Dang v. Cross, 422 F.3d 800, 811 (9th Cir. 2005); see also BladeRoom Group Ltd. v. Emerson Electric Co., 20 F.4th 1231, 1243–44 (9th Cir. 2021) (rejecting a universal “tie-to- the-verdict” rule). We have explained that prejudice is likely where the error adds an element to a party’s burden of proof and where “nothing about th[e jury’s] verdict indicates that the result would have been the same without the error.” Clem, 566 F.3d at 1182–83 (alteration in original), quoting Caballero v. City of Concord, 956 F.2d 204, 207 (9th Cir. 1992). This is the general rule because “the burden of proof affects all aspects of the jury’s verdict,” and an error can make it “impossible to determine whether the erroneous 5 This is a case where “counsel shifted his position as advocate precisely because” of an erroneous ruling. See United States v. Espinoza- Bazaldua, 711 F. App’x 737, 741 (5th Cir. 2017) (non-precedential) (rejecting argument that defendant invited the error he argued on appeal).  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 15 burden of proof was outcome determinative.” Carvalho v. Raybestos-Manhattan, Inc., 794 F.2d 454, 455 (9th Cir. 1986). Comet offers two grounds for finding this instructional error harmless: (1) that the instructions were accurate as a whole; and (2) that Comet’s evidence was so strong that the jury probably would have ruled the same way even if it had been properly instructed. We reject both contentions. 1. Accuracy as a Whole A single mistaken instruction may be deemed harmless if the instructions, taken as a whole, fairly and correctly cover the substance of the applicable law. Swinton v. Potomac Corp., 270 F.3d 794, 807 (9th Cir. 2001). Comet points out that Instruction 18, a separate instruction, correctly told the jury that Comet bore the burden of proof on another element of its claims: that the alleged trade secrets gave Comet “independent economic value” through providing a “business advantage.” Comet contends that by satisfying its burden on this element under Instruction 18, it necessarily proved that its alleged trade secrets were not readily ascertainable by proper means. On this subject, the instructions were not accurate as a whole. Instruction 20 flipped the burden of proof on lack of ready ascertainability. Instruction 18 addressed utility in terms of “independent economic value,” while “ready ascertainability” addresses secrecy. Information can be secret without being useful or useful without being secret. The two may often go together, but neither feature necessarily implies the other. For three reasons, we reject Comet’s argument, which the dissent seems to adopt, that by proving a trade secret 16 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC provides independent economic value, it proved “as a matter of logic” that the information was not readily ascertainable. First, that position is undermined by Comet’s admission in the district court that “Instruction No. 18 . . . omits [the lack of ready ascertainability] aspect of the trade-secret definition under the DTSA,” a point it made when arguing to remove Instruction 20 altogether. Second, and more fundamental, Comet’s theory would effectively rewrite the DTSA, which defines a trade secret to require both independent economic value and lack of ready ascertainability as related but distinct statutory elements. See 18 U.S.C. § 1839(3)(B) (“[T]he information derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable through proper means by, another person who can obtain economic value from the disclosure or use of the information.”). Third, a reasonable juror would easily conclude that, because Instructions 18 and 20 were separate and tied to separate questions on the verdict form, the two required independent consideration. 2. Likely Result The party defending the judgment bears the burden of showing that the jury more likely than not would have reached the same result if it had been properly instructed. Clem, 566 F.3d at 1182. Comet must either: (1) point to something about the jury’s verdict from which we can infer how it would have ruled absent the mistake; or (2) convince us that its evidence was so strong that we can step confidently into the jury’s shoes to make that determination ourselves. See Mockler v. Multnomah County, 140 F.3d 808, 812–14 (9th Cir. 1998) (making inferences from jury’s verdict and reviewing weight of evidence in addressing harmlessness). Comet has done neither.  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 17 Nothing in the jury’s complete verdict sheds light on how it would have ruled if properly instructed on lack of ready ascertainability. To be sure, the jury found that XP misappropriated some of the alleged trade secrets and did so willfully and maliciously, and the jury awarded compensatory and punitive damages. Those findings were all supported by sufficient evidence, but sufficiency is not enough to show an instructional error was harmless. Lack of ready ascertainability under the DTSA has nothing to do with the defendant’s real-world conduct. The jury had to decide whether XP or others could have reverse-engineered the alleged trade secrets, not whether XP had actually done so. On the ability to reverse-engineer, the jury heard sharply conflicting evidence, which we review below. As to whether XP could have reverse-engineered the alleged trade secrets, the jury’s finding about what actually happened does not tell us what the jury would have decided—under a correct instruction on the burden of proof—about what XP could have theoretically done to reverse-engineer any of the information. These two aspects of the jury’s verdict are distinct. One asks what XP did and the other what XP and others could have done. The award of compensatory damages for “the value of XP’s benefit that would not have been achieved except for their misappropriation,” as required by Instruction 23, is a red herring on the harmless-error question. XP did not argue that it would have reverse-engineered the alleged trade secrets if not for acquiring them through Mr. Mason, merely that it could have done so, at least in part. Likewise, the jury’s award of punitive damages does not support harmlessness on lack of ready ascertainability, as the dissent asserts. In past cases noting a punitive damages award as a significant factor in finding harmlessness, the aspect of the 18 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC case on which the jury was improperly instructed bore a direct logical relationship to the findings required to award punitive damages. See Larez v. Holcomb, 16 F.3d 1513, 1518 (9th Cir. 1994) (jury’s finding that police officer engaged in “extraordinary misconduct” warranting punitive damages was “highly significant” in gauging how a properly instructed jury likely would have ruled on a consent-to-arrest defense); Lambert v. Ackerley, 180 F.3d 997, 1008–10 (9th Cir. 1999) (en banc) (failure to give instruction on affirmative defense of economic restructuring was harmless where jury awarded punitive damages and therefore must have found defendants acted with a culpable mens rea). Here, however, ready ascertainability or its absence did not depend on whether XP engaged in egregious misconduct, of which there was ample evidence. Instead, the jury had to weigh conflicting testimony from experts on what XP or others could have reverse-engineered. In general, weighing conflicting expert testimony falls outside the wheelhouse of a federal court of appeals. See Wyler Summit P’ship v. Turner Broad. Sys., Inc., 235 F.3d 1184, 1192, 1196 (9th Cir. 2000) (reversing summary judgment in case of dueling experts because neither side’s theory was barred as a matter of law). This is not a case where we could simply apply familiar legal standards to undisputed facts about who said and did what when. See Mockler, 140 F.3d at 812–14 (affirming jury verdict despite instructional error in employment discrimination case; panel could conclude remedial action was inadequate from unrebutted factual testimony). Rather, we must review highly technical expert testimony and do so mostly without the benefit of access to the many exhibits those witnesses used to illustrate and explain their testimony. We do not agree with Comet that its evidence on lack of ready ascertainability was so overwhelming that we should  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 19 deem the error harmless. The first major obstacle to a finding of harmless error here is Comet’s attempt to rebut the testimony of XP’s key expert witness, Dr. Joshua Phinney. It was a study in changing the subject. In cross-examination of Dr. Phinney, Comet understandably emphasized the defendants’ bad acts in taking and using confidential information from Comet. But the cross-examination does not seem to have laid a glove on Dr. Phinney’s points about significant aspects of the alleged trade secrets actually consisting of information that could have been reverse- engineered or was otherwise publicly disclosed. Similarly, Comet’s closing argument on lack of ready ascertainability amounted only to pointing out that “XP kept going back to the Comet documents. There’s no evidence that XP went out and looked publicly for this stuff.” 6 At the risk of some repetition, lack of ready ascertainability does not depend on what the defendants actually did. Ready ascertainability or its absence, as here, depends on an evaluation of what another expert in the field could have learned by proper means. Comet’s failure to defang Dr. Phinney’s testimony on this subject weighs against a finding of harmless error. Even if we could say with some confidence that some of Comet’s alleged trade secrets were genuinely protected, meaning that the information was not readily ascertainable with legitimate methods, there are further obstacles to finding harmless error. The jury found in favor of XP on 6 Perhaps Comet focused on XP’s conduct when cross-examining Dr. Phinney because at the time it was still asserting claims under the California UTSA, for which ready ascertainability is an affirmative defense. It requires the defendant actually to have obtained the information lawfully. Supra n.4. 20 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC Trade Secret S, and it found that Trade Secret L was misappropriated but that Comet did not prove damages for Trade Secret L. Further, the compensatory damage award was $5 million for Trade Secret D and $15 million for Trade Secret E. The combination of verdict answers makes it impossible to tease out how a failure to prove lack of ready ascertainability as to some elements of the claimed trade secrets would affect the compensatory damage award. Even if we assume for the sake of argument that some of the documents and computer files identified by Comet and our dissenting colleague contained at least some information that was not readily ascertainable, we could not find the error was harmless. Comet’s damages were based on its own total research and development expenses. In particular, Comet’s damages expert used internal cost codes to calculate development costs for each alleged trade secret—a common method of intellectual property valuation—and “assume[d] that every bit of confidential information included in Comet’s prior generation product codes . . . was valuable to XP.” He relied on other witnesses to identify these cost codes and lacked the technical expertise to opine that they reflected only protected elements of the alleged trade secrets. On appeal, Comet provides no basis for identifying which of those expenses were implicated by various sorts of information that may have been readily ascertainable by proper means, such as through reverse-engineering, by XP or others. At an absolute minimum, XP correctly identifies, the jury would have had to award a smaller amount in compensatory damages if it concluded that certain aspects of the alleged trade secrets were readily ascertainable and thus not protected by the DTSA. Given the conflicting testimony on the question of lack of ready ascertainability, which we review below, we cannot be confident that the instructional  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 21 error was harmless as to all of the stolen information. Comet all but conceded as such when it wrote that “Trade Secrets D and E encompassed a vast library of information, much of which could not be obtained from reverse engineering.” Appellee’s Br. at 32–33 (emphasis added). Comet characterized Dr. Phinney’s testimony as being that the “dimensions and configuration” of similar components in competitors’ products could be replicated by opening them up and that public patents disclosed certain “concept[s]” used in the alleged trade secrets. Id. at 35–36. Indeed, XP introduced testimony intended to show that the alleged trade secrets were readily ascertainable “through various means, including by opening a match box and reviewing public patents.” Appellant’s Br. at 25. For example, Dr. Phinney testified that significant aspects of Comet’s alleged trade secrets were disclosed in other companies’ public patents and product bulletins. Anyone wishing to replicate the physical dimensions of a product could easily measure them, Dr. Phinney testified, and in his opinion certain components used a “standard format or form factor” and were derived from “off-the-shelf” parts. In response to Dr. Phinney’s testimony, Comet contends that “other vital aspects” of its trade secrets, such as “manufacturing variation and material properties,” could not have been learned through reverse-engineering. Appellees’ Br. at 35–36. Comet also argues that the patents Dr. Phinney relied upon did not reveal the “strategy behind design choices, or how to implement the technology.” Id. at 36. For example, Dr. Shanfield testified that proprietary performance test data the engineers brought from Comet to XP was “the key” to choosing the right components. Comet also noted that its products containing Trade Secrets D and 22 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC E were not yet available for purchase at the time of the theft, making reverse-engineering anything from them impossible. However, Dr. Phinney’s testimony largely concerned public information about other companies’ products already on the market. By his account, skilled electrical engineers would easily have understood Comet’s design strategy. For example, Dr. Phinney identified one aspect of Trade Secret E as “a component that has been used for a long time in this industry,” one that “everyone knew” was needed. He also criticized Comet’s expert, Dr. Shanfield, for failing to address what was known generally in the industry and for failing to engage on many of the details of the alleged trade secrets so as not to “bore the jury.” Critical too, Trade Secret L was already on the market. Comet itself wrote in another section of its brief that the “award of $0 in damages for Trade Secret L was given in tandem with an award for Trade Secret E that incorporated development costs for Trade Secret L, reflecting an effort to avoid double counting.” Appellees’ Br. at 21–22. That is, Comet had asked for $6 million for Trade Secret E and $11.1 million for Trade Secret L, and eventually won a $15 million award for Trade Secret E. Accordingly, on Comet’s own theory of the case, the development costs for Trade Secret L must have been the majority of the award for Trade Secret E, and close to (if not a majority) of the entire compensatory damages award of $20 million. In Comet’s own words: “The jury’s damages awards for Trade Secrets E and L are thus intertwined and cannot fairly be reconsidered separately.” Id. at 22. We simply cannot say on this record how much the jury would likely have awarded for research and development costs going into “manufacturing variation,” “material  COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC 23 properties,” and “strategy behind design choices,” after excluding costs going into the potentially readily ascertainable categories of “dimensions” and “configuration,” as well as certain “concepts.” See BladeRoom, 20 F.4th at 1246 (vacating damages award and noting court has “no way to know how much damages the jury intended to allocate for each claim” and “cannot tell whether the damages findings would be the same”). Moreover, even if we were to ignore the patents as offered at trial solely for secrecy and not ready ascertainability, as Comet and the dissent insist, some significant portion of the design files Mason stole must reflect “dimensions” and “configuration.” One Comet witness described the “documents that you create as part of your design process” as starting with “specification and concepts” and then moving on to “schematics, to layouts; for example, for PCB boards, 3-D drawings.” Comet chose to propose a bottom-line damages figure for each alleged trade secret rather than risk boring or confusing the jury with a more detailed component-by- component or document-by-document breakdown. With the benefit of hindsight, that understandable strategic choice created obstacles to a finding of harmless error. If the jury had arrived at the same verdict after being properly instructed, we would have applied much more forgiving standards of review on appeal. See Acosta v. City of Costa Mesa, 718 F.3d 800, 828 (9th Cir. 2013) (per curiam) (liability); Oracle Corp. v. SAP AG, 765 F.3d 1081, 1094 (9th Cir. 2014) (remittitur). The conflicting evidence on the many issues of ready ascertainability and the ensuing damage calculations presented the sorts of disputes a properly instructed jury must decide, so we cannot overlook the instructional error as 24 COMET TECHNOLOGIES USA, INC. V. XP POWER, LLC harmless. We therefore vacate the district court’s amended final judgment in its entirety, including the permanent injunction, though the jury’s verdict that XP did not misappropriate Trade Secret S and the grant of judgment as a matter of law for XP on Trade Secret T are unchallenged on appeal and thus rema