WASHINGTON, May 7, 2026, 17:06 (EDT)
The Pentagon has boosted its artificial intelligence pact with Meta-backed Scale AI, raising the contract ceiling to $500 million—a fivefold increase that hands the startup more responsibility in sifting data and bolstering military decision-making. Bloomberg Law noted this follows Scale’s earlier $100 million award from September.
This shift has teeth: the Pentagon is pushing AI out of test labs and into systems military units actually use. Washington Technology says the expanded deal includes computer vision—AI trained to interpret images—and generative AI, which pulls data together or drafts material for users.
Just days ago, the department announced it had struck classified-network deals with leading AI and cloud players—Google, OpenAI, Microsoft—aiming to roll out advanced technology for what it described as “lawful operational use.” The Pentagon said over 1.3 million personnel have tapped its GenAI.mil platform in five months. U.S. Department of War
Dan Tadross, head of Scale AI’s public-sector division, told Bloomberg Law the Pentagon had been “pushing the limits” with its previous agreement. “I think this contract is just generally proof that the department is eager to adopt this technology,” Tadross said. Bloomberg Law
Scale described the deal as a Production Other Transaction Authority agreement—essentially a speedier tool for Pentagon procurement, where units can strike project agreements via one central office instead of opening new competitions. According to the company, demand already blew past the initial cap, with work ranging from computer vision and generative AI decision support to data operations.
Scale is putting its Data Engine, GenAI Platform, and Donovan—the decision-support tech aimed at defense and intelligence teams grappling with big sets of unstructured data—on the table as part of the deal. According to GovCon Wire, these offerings are intended for model testing, deployment, and bringing operational visibility across both classified and unclassified networks.
This isn’t just a race to build the largest model. According to The Next Web, Google, OpenAI, and Microsoft are entrenched in the Pentagon’s classified AI efforts at the model-and-cloud level. Scale, on the other hand, has carved out a niche higher up the stack, focusing on supplying data and decision-support tools — the kind military operators use to prep information for AI systems.
Those deals have stirred up familiar disputes within Silicon Valley. Google has moved forward on Pentagon AI projects, even as staff push back. Company spokeswoman Jenn Crider said Google was “proud” to be part of the larger team delivering AI and cloud services for U.S. national security. She added the company stays against domestic mass surveillance or letting autonomous weapons operate without a human in the loop. The Independent
Employees at Google DeepMind’s London office have called on management to grant formal recognition to the Communication Workers Union and Unite the Union, according to WIRED. John Chadfield, national officer for technology at the CWU, described the move as a way to hold Google accountable to its stated AI principles—and to boost workers’ bargaining power in the face of “an increasingly deaf management.” WIRED
Google pushed back on some of those details. According to The Verge, company spokesperson Kristen Morea confirmed receipt of a union-recognition letter but clarified, “at this stage in the process, there has been no vote to unionise.” Morea added that Google values constructive dialogue with its employees. The Verge
There’s a risk here: spending could easily outpace governance. Setting a $500 million cap doesn’t mean that money will be spent, but as Google’s recent internal conflicts make clear, contract language built on “lawful” use and human oversight doesn’t erase anxieties among workers, investors, or the public about AI’s role in surveillance, targeting, or military operations. More than 600 Google employees signed an open letter—reported by The Guardian—calling on the company to keep its AI out of classified applications. The Guardian
Right now, the Pentagon is prioritizing fast options and flexibility. According to Al Jazeera, officials described their latest AI deals as an effort to dodge “vendor lock”—the risk of getting stuck with a single provider—while also equipping troops with new tools for processing information and improving situational awareness. With Scale’s bigger contract in play, it’s clear that competition isn’t just about the AI models anymore; control over clean data and decision-support tech is just as fierce. Aljazeera
