In March 2022, the Department of Justice sentenced JHL Biotech co-founders Racho Jordanov and Rose Lin for conspiracy to steal Genentech trade secrets and commit wire fraud exceeding $101 million. The defendants had funneled proprietary biopharmaceutical data to their Taiwan-based competitor by recruiting former Genentech employees, enabling biosimilar development that prosecutors said allowed JHL to “cheat, cut corners … and avoid further experimentation,” according to the Department of Justice. Yet the real mechanism enabling prosecution was resource-based filtering, not effective protection. Genentech possessed the documentation infrastructure to identify what was taken and the litigation budget to prove it. Most biotech and AI startups possess neither.
This dynamic reflects what practitioners term the “infrastructure gap:” the disparity between trade secret law’s documentation requirements and early-stage companies’ operational capacity. The Defend Trade Secrets Act and state Uniform Trade Secrets Act implementations require “reasonable measures” to maintain secrecy, while many courts and some state statutes separately require plaintiffs to identify alleged trade secrets with “reasonable particularity” early in litigation. Courts often consider classification schemes, access controls, confidentiality markings, and documented exit procedures as evidence of protective measures. The WIPO Guide recommends internal registries and systematic access restrictions. A well-resourced company implements these systems from incorporation. A three-person biotech team cannot.
The dependence on trade secrets reflects structural necessity, not strategic preference. Industry analysis indicates early-stage companies avoid patents because disclosure reveals competitive intelligence, examination timelines exceed development cycles, and enforcement costs surpass acquisition budgets. Recent studies report that AI innovation now outpaces patent examination entirely. China’s documented targeting of intellectual property in biotech and semiconductors compounds the pressure to rely on trade secrets rather than patents. For most founders, secrecy is the only viable protection for innovations too early to patent and too valuable to disclose.
The legal framework’s design offered apparent advantages: protect secrets without freezing talent. The DTSA includes an explicit mobility provision prohibiting courts from issuing orders that “prevent a person from entering into an employment relationship.” California’s Section 16600 voids most employment non-compete agreements. Kewanee Oil confirmed that state trade secret protection can coexist with the federal patent system. The resulting doctrine channels remedies toward artifacts–meaning specific files, formulas, and documented processes–rather than employment restrictions. Yet the implementation created a filtering mechanism that eliminates under-resourced plaintiffs before substantive evaluation.
In Double Eagle Alloys v. Hooper, the employer alleged a former employee took 2,660 files and described its secrets in broad categories: “specifications, pricing, margins, costs, and customer drawings.” The Tenth Circuit affirmed summary judgment for the defendant largely because Double Eagle could not identify and distinguish its alleged trade secrets with enough specificity to proceed. Theft was not disproven; the company simply could not articulate what qualified as secret. The Sedona Conference defends early identification requirements as improving notice, efficiency, and preventing overbroad discovery. Yet the requirement filtered out a company with legitimate grievances and inadequate systems to describe them.
Empirical data confirms this pattern. While one study reports that in a representative sample of trade secret cases filed between 2017 and 2022 that resulted in judgment, plaintiffs received favorable outcomes in 81% of cases, and among the subset that proceeded to federal trial, judgment favored claimants roughly 84% of the time. Yet these numbers measure only companies sophisticated enough to survive threshold challenges and wealthy enough to reach judgment. Separate analysis of cases from 2009 through 2018 found 11% dismissed specifically because plaintiffs failed to demonstrate reasonable protective measures. Documentation failures disqualified claims before courts examined whether misappropriation occurred.
The jurisdictional landscape compounds these disadvantages. PepsiCo v. Redmond established inevitable disclosure in the Seventh Circuit, permitting injunctions when employees would inevitably rely on former employers’ secrets. The court reasoned the defendant could not possess “an uncanny ability to compartmentalize information.” California rejected this doctrine outright. Whyte v. Schlage held that inevitable disclosure “creates an after-the-fact covenant not to compete.” The same departure triggers relief in Illinois and none in California. Sophisticated companies exploit this divergence through forum selection and choice-of-law provisions. Startups lack the legal infrastructure to navigate jurisdictional strategy.
The framework designed to protect mobility ultimately undermines it for companies that need talent most. When startups cannot enforce trade secret claims, researchers face a calculus: join an established company with documented infrastructure and enforceable rights, or join a startup where a co-founder’s departure could leave innovations unprotected. The Waymo-Uber settlement, which resulted in approximately $245 million in Uber equity plus commitments regarding non-use of Waymo confidential information, succeeded because Waymo could identify classified, documented artifacts. A startup without equivalent systems cannot offer equivalent assurance. The protection gap does not merely disadvantage founders in litigation. It disadvantages them in recruiting.
Startups can narrow the infrastructure gap through targeted practices. Marking documents as “confidential” at creation rather than retroactively cataloguing creates contemporaneous evidence. Documenting access controls in onboarding emails, even informal ones, builds a record of protective measures. Exit interviews that identify specific files accessed establish artifact trails. These measures protect trade secrets through documentation rather than mobility restrictions, preserving talent fluidity while building evidentiary foundations. The goal is sufficient documentation to survive threshold challenges, not comprehensive security infrastructure.
Trade secret law protects what companies can document; it cannot protect what companies cannot yet articulate. Particularity requirements, documentation standards, and artifact-based remedies do not discriminate based on innovation value or theft severity. They discriminate based on resources. Recalibrating particularity standards to account for operational capacity and permitting reasonable inference from circumstantial evidence when direct documentation is unavailable would better align trade secret protection with the innovation ecosystem it purports to serve. Creating safe harbors for early-stage documentation practices or scaling “reasonable measures” requirements to company size would preserve mobility while ensuring the framework protects those with the most to lose, not merely those with the most to spend.