Scale AI has sued a former employee and Mercor, a fast-emerging competitor in AI data operations, claiming an effort to poach some of Scale’s biggest enterprise customers through misuse of confidential materials. The lawsuit accuses Mercor of misappropriating trade secrets, and alleges that the ex-staffer, Eugene Ling, violated contractual obligations while he was still on Scale’s payroll.
‘Customer A’ and strategic documents at heart of allegations
Central to the filing is an unnamed “Customer A,” whom the complaint identifies as one of Scale’s largest accounts. Ling began pitching Mercor as a potential vendor to this customer already before his departure, Scale says, and he kept internal files about customer strategies, pricing, work plans and delivery playbooks. The company says such materials would provide Mercor not only with a roadmap to serve Customer A, but also for several other well-known clients — contracts that Scale says could be worth millions of dollars to a rival.
Trade secrets lawsuits in enterprise data services are often about whether the contested matter is actually a confidential competitive asset, and, if so, then did it translate to commercial gain. Things on such issues as how specific the allegedly secret information is, what efforts the employer is making to keep it secret and when the applicant is doing the competitive solicitation.
Mercor and the ex-employee fight back
Mercor co-founder Surya Midha has denied that his company misappropriated Scale’s data, and says he believes Ling plugged into Scale before his defeat in that year’s Synergy planning competition and potentially retained files after he left. He added that Mercor was open to cooperating, including on procedures of deletion or other, and was waiting for guidance from Scale. Ling said in separate comments that he never used the materials in his new capacity, characterized the retention as unintentional and suggested that he was waiting for direction on how to deal with the situation.
And these answers will probably affect any early judicial decisions around injunctive relief. Judges often consider whether the accused party took prompt steps to retain, sequester, and return contested files and whether there are plausible reasons for having them.
Why this fight matters in the AI vendor ecosystem
The stakes are high because big AI data contracts are rare in number and stickiness. Public disclosures from Appen, one of just a handful of data-annotation firms that are listed, illustrate how one hyperscale customer, which can account for a 10 percent slice of annual revenue, would make any decision to switch vendors hugely impactful. More generally, Gartner and IDC have observed that enterprise AI spend is still concentrated with a small group of tech buyers, which increases the value of each flagship account for service providers like Scale and Mercor.
The competitive landscape is further heating up due to the proliferation of foundational models and the rising need for better-quality domain-specific data. Mercor has positioned itself as a high-end vendor by harnessing subject-matter experts — usually PhDs — to curate and evaluate training data for large language models. That approach is counter to the old ways of crowdsourcing and is representative of the rising focus on quality, safety, and evaluation that the Stanford reports touch on.
Perceptions of conflicts after high-profile partnerships
Industry reporting over the past several months has shed a spotlight on concerns over vendor neutrality in the AI stack. The latter emerged, in media reports, as Scale’s strengthening relationship with a large platform company caused some to re-evaluate their business, large data customers that compete with that platform. The mere hint of a conflict can drive companies to run parallel pilots or rebid work, raising competitive pressure from challengers like Mercor, Labelbox, Appen and Sama.
That might also explain why a dispute with a single customer can send a ripple through the market. If Scale’s assertions are found valid by the courts, competitors could come under closer scrutiny of employee onboarding and data hygiene. If the allegations don’t hold up, it could embolden customers to diversity vendors more aggressively — the worst fear of tech companies facing litigation — without getting caught in the legal crossfire.
What courts may now be looking for
Early stages often entail temporary restraining orders, forensics examinations of devices and cloud accounts and the preservation of evidence. Subpoenas for “Customer A” might confirm whether or not any Scale materials had been used to form competing bids or delivery plans. Courts also frequently parse the boundaries and enforceability of non-solicitation and confidentiality covenants, particularly in jurisdictions such as California that ban non-competes but recognize trade secret protections.
For customers, the pragmatic takeaway is simple: Demand clean-room operations, documented handoffs and hard walls between competitive vendors. For AI data providers, the moment is proof of the need for tight access controls, automated offboarding and auditable retention policies—areas many organizations say they have bolstered as model deployment has grown.
Bottom line
Scale AI’s lawsuit sets up a high-stakes battle for control over the playbooks of how to serve the most powerful buyers of A.I. Both Mercor and the ex-employee deny that abuse has happened, and indicate a willingness to remedy any problems. The result will determine not only the relationship between the two companies and a critical group of accounts, but also the guardrails between hiring and competItion in the lucrative, consolidating market for AI training data and evaluation services.
