AMD v. Feldstein: Trade Secrets | Outside IP Counsel

A District of Massachusetts judge enjoined four former employees who walked to Nvidia after three of them copied AMD files to personal drives, one of them more than a million, holding that improper acquisition, not proven use, supports a trade-secret injunction.

The forensic record is what makes Advanced Micro Devices, Inc. v. Feldstein a teaching case. Judge Timothy S. Hillman of the U.S. District Court for the District of Massachusetts granted AMD a preliminary injunction in an unreported memorandum dated May 15, 2013 (Docket No. 105) in Civil Action No. 13-40007-TSH. The reported opinion in the case, 951 F. Supp. 2d 212 (D. Mass. June 10, 2013), is the later ruling on the defendants’ motions to dismiss, which recites the same facts and is best known for its Computer Fraud and Abuse Act analysis. Both arose after AMD employees decamped, over a period of months, for the competitor Nvidia. Four defendants were enjoined (Robert Feldstein, Manoo Desai, Nicolas Kociuk, and Richard Hagen); a fifth, Deepaksrivats Thirumalai, was added by a second amended complaint in March 2013 and drew a temporary restraining order. The departures were not quiet. In the days before they resigned, three of them (Feldstein, Desai, and Kociuk) copied AMD files to personal external storage devices on a scale that turned an ordinary departure into a misappropriation suit. Kociuk alone copied more than a million files. That number, more than any single document, is the reason the case is taught.

At a glance

Case: Advanced Micro Devices, Inc. v. Feldstein, No. 13-40007-TSH (D. Mass. May 15, 2013) (preliminary injunction, unreported); companion reported opinion on the motions to dismiss at 951 F. Supp. 2d 212 (D. Mass. June 10, 2013)
Court: U.S. District Court for the District of Massachusetts (Worcester), Civil Action No. 13-40007-TSH
Judge: Timothy S. Hillman, U.S. District Judge
Posture: AMD obtained a temporary restraining order in January 2013; the May 15, 2013 memorandum resolved AMD’s motion for a preliminary injunction
Claims: Seven counts: misappropriation of trade secrets under Massachusetts common law and under Mass. Gen. Laws ch. 93, §§ 42 and 42A; unfair competition under ch. 93A, § 11; violation of the federal Computer Fraud and Abuse Act, 18 U.S.C. § 1030; breach of the employee’s duty of loyalty; breach of contract; and conspiracy
Core holding: A plaintiff need not prove that a departing employee actually used or disclosed a trade secret to obtain a preliminary injunction; the improper acquisition of trade secrets, with intent to convert them, can itself support injunctive relief and a likelihood of success on the merits
Result: Preliminary injunction granted against four defendants, restraining use or disclosure of AMD’s confidential information and reaching solicitation of AMD employees. AMD’s solicitation claim against Feldstein was the one exception: on that claim it failed to meet the standard for injunctive relief

The exfiltration that defined the record

Most departing-employee trade-secret cases are won or lost on circumstantial inference: an unusual overlap in a competitor’s product, a suspiciously quick launch, a customer who jumps. Feldstein is unusual because the central facts were not inferred at all. They were logged. Each of the three employees who copied files left a device-level trail, and the aggregate volume was staggering. AMD’s January 2013 complaint alleged that the defendants had collectively transferred more than 100,000 electronic files to storage devices. By the preliminary-injunction stage the forensic picture was more specific, and far larger: 8,148 files copied from AMD’s intranet by way of Feldstein’s AMD-issued laptop during a sabbatical he took shortly before resigning; 7,899 documents moved from Desai’s AMD laptop to a folder on an external hard drive; and, from Kociuk, more than a million files copied onto a pair of external hard drives. Kociuk’s copy was the detail that collapsed the defense’s narrative. He acknowledged using the Robocopy utility to duplicate the entire file structure of both of his AMD-issued computers, explaining that he wanted to preserve personal data. A person who genuinely intends to take only what is lawfully his does not image what amounts to a substantial slice of an engineering department.

Among the materials were files AMD identified as plainly proprietary: Feldstein moved to external storage three highly confidential files, including two licensing agreements with significant customers and a document setting out proposed strategy for AMD’s technology licensing. He also downloaded a “Technology Licensing Overview” presentation that he later conceded was problematic. The presence of specific, high-sensitivity documents mattered, because it answered the predictable defense that the copying was indiscriminate digital clutter: old personal files, public materials, and the detritus that accumulates on any work machine. The combination of scale and specificity is what gave the record its force: not merely a lot of files, but a lot of files that included exactly the kind of competitively sensitive material a rival would want.

For litigators, the lesson is procedural as much as substantive. AMD moved quickly, secured a temporary restraining order, and preserved the devices before the story could be tidied up. The forensic window around a resignation is narrow and fragile; metadata can be overwritten, drives can be wiped, and “I deleted it all, I never looked at it” becomes harder to disprove with every passing week. Feldstein rewards the plaintiff who treats the first forty-eight hours after discovery as an evidence-preservation emergency.

Acquisition through improper means, not proof of use

The doctrinal heart of the opinion is its treatment of what a trade-secret plaintiff must show at the preliminary-injunction stage. The defendants pressed the intuitive argument that AMD could not point to a single instance in which any of them had actually used an AMD trade secret at Nvidia or disclosed one to their new employer. No use, they urged, meant no irreparable harm and no likelihood of success.

Judge Hillman rejected the premise that actual use is a prerequisite. Massachusetts trade-secret law, like the Uniform Trade Secrets Act framework it tracks, defines misappropriation to include the acquisition of a trade secret by improper means, separate and apart from its later use or disclosure. On that understanding, an employee who copies an employer’s protected files onto personal media without authorization, intending to carry them to a competitor, has already done something the law forbids; the plaintiff need not wait until the secret surfaces in a rival’s product to seek relief. The court framed the operative question as whether AMD could show the acquisition, through improper means, of trade secrets with the intent to convert them to the use of someone other than the rightful owner. On this record, it could.

This is the holding practitioners cite. It reorients the irreparable-harm inquiry away from the hard-to-prove fact of present use and toward the demonstrable fact of improper acquisition plus the risk of future use. That reorientation is powerful precisely in the employee-mobility setting, where the new employer and the departed engineer are the only people who know whether and how the files have been touched, and where waiting for proof of use often means waiting until the harm is irreversible.

The line between portable skill and the company’s files

Every departing-employee case has to confront the principle that an employee may carry general skill, knowledge, and experience to a new job. Engineers who spend years at AMD learn how to design competitive graphics and computing architectures, and that learning is theirs to sell on the open market. Nothing in Feldstein questions that. What the case illustrates is how decisively a documentary exfiltration record removes the dispute from that protected zone. The defendants were not accused of remembering too much; they were shown to have copied the company’s actual files. General knowledge lives in a person’s head and travels with them lawfully. A million files on a personal drive are not knowledge. They are the employer’s property, and copying them is an act, not a memory.

The opinion also touched the related problem of employee solicitation. AMD’s theory was not only that the employees took files but that the departures were coordinated, with earlier movers helping recruit later ones. The injunction accordingly reached solicitation of AMD personnel, reflecting the reality that a coordinated lift-out of a team can do competitive damage independent of any single document. The theory was not uniformly successful, though. The court found AMD likely to succeed on its contractual solicitation claims against Desai and Hagen, but not against Feldstein, where the evidence went little further than his making positive comments about Nvidia. That claim survived a later motion to dismiss under the lower Rule 12(b)(6) threshold, but it did not support an injunction.

Open questions

The opinion leaves the contours of “improper acquisition” relief less than fully mapped. How much copying is enough? Does a handful of misdirected files support the same inference of intent as a million? The case answers the easy end of that spectrum and leaves the middle to judgment. The court’s willingness to enjoin on acquisition-plus-risk also sits near the contested boundary of threatened misappropriation; courts remain divided about how far a plaintiff may go in restraining a competitor’s new hire before the injunction functions as a de facto noncompete. And because Feldstein resolved a preliminary-injunction motion rather than a final judgment, it fixes a likelihood of success, not an ultimate adjudication that any particular file was a trade secret.

Implications

Volume is an argument. A massive, indiscriminate copy of company files in the days before departure is hard to reconcile with any innocent explanation and can carry a misappropriation showing on its own.
Acquisition can be the violation. Under the UTSA framework applied here, improper acquisition with intent to convert supports relief; a plaintiff need not prove the secret was already used.
Move fast and preserve. AMD’s early TRO and device preservation locked in the forensic record before it could degrade. That is the decisive litigation choice in exfiltration cases.
Identify the crown jewels. Pointing to specific high-sensitivity documents within the larger haul rebuts the “it was all junk” defense and shows real competitive risk.
Solicitation travels with the files. Where departures are coordinated, an injunction may reach recruitment of remaining employees, not just use of documents.

Frequently asked questions

Did AMD have to prove the engineers gave anything to Nvidia? No. The court held that AMD did not need to show actual use or disclosure to obtain a preliminary injunction. The improper acquisition of trade secrets, coupled with intent to convert them, was enough to support a likelihood of success and injunctive relief.

Was the case about what the engineers remembered, or what they copied? What they copied. Employees may lawfully carry general skill and experience to a competitor. Feldstein turned on the copying of the employer’s actual files to personal devices (conduct, not memory), which is why the general-skill defense had little purchase.

What role did the Computer Fraud and Abuse Act play? AMD pleaded a CFAA claim alleging unauthorized or excess-authorized access to its protected network, alongside its trade-secret and contract claims. In the reported June 10, 2013 opinion on the motions to dismiss, Judge Hillman sided with the narrower, technological reading of “authorized access,” under which logging in with valid credentials is authorized even if the employee later misuses what he obtained. He declined to dismiss the CFAA count anyway, given the unsettled law and the incomplete record, but warned that the claim would be dismissed unless AMD could plead fraud, deception, or the circumvention of technological barriers. That CFAA discussion is what the reported opinion is most cited for.