Anthropic, Google, OpenAI, and xAI get $800M to hop in bed with Pentagon

Looks like DoD FOMO struck Silicon Valley

Updated The Pentagon's embrace of the AI industry just put up to $800 million on the table, as the Department of Defense issued a quartet of contracts bringing the biggest names in the biz officially into the fold. 

The Pentagon's Chief Digital and Artificial Intelligence Office (CDAO) today announced contracts with Anthropic, Google, OpenAI, and xAI to help it use artificial intelligence for a multitude of military purposes - each with a $200 million ceiling. 

The four firms contracted for the work are all behind AI models the DoD believes have game-changing defense applications. The Pentagon appears particularly eager to pursue agentic AI, mentioning in the announcement that it hopes to use the skills of AI experts at the four firms "to develop agentic AI workflows across a variety of mission areas." 

The DoD hopes the early experiments will help it understand how to apply AI for its purposes, while the four firms will also be asked "to understand and address critical national security needs." 

"Leveraging commercially available solutions into an integrated capabilities approach will accelerate the use of advanced AI as part of our Joint mission essential tasks," said CDAO Chief Digital and AI Officer Doug Matty. 

While the benefits for the four contractors are fairly obvious, the DoD is also getting a good deal, it thinks. 

America's military is getting access to Google's newest AI models, for example, and the Chocolate Factory is also offering the Pentagon access to Google infrastructure around the country. Anthropic, meanwhile, said that it will be helping the military identify the best AI use cases and developing models tuned on DoD data, identifying and mitigating adversarial uses of AI, and working to "prototype frontier AI capabilities that advance US national security."

Anthropic did note that its multi-million-dollar contract with the DoD is for two years, answering one of several questions we asked that the Pentagon declined to answer. 

Grok führer Government

In an abysmal example of timing, xAI took today's contract announcement as a chance to announce its own new product: Grok for Government, "a suite of frontier AI products available to US Government customers."

No word on whether they'll come equipped with MechaHitler mode, but there's not a lot of word on how xAI intends to stand out from the pack in the release, either. According to xAI, Grok for Government means all its "world-class AI tools" will be available to public sector workers, and various standard software helpers are on offer, too.

With controversy around Musk a daily occurrence, and Grok less than reliable at best, it remains to be seen whether this move will earn Musk more money, or actually do any good. 

xAI, meanwhile, only noted the contract along with a deal that now sees its products available for purchase by the US via the General Services Administration, meaning they're generally available across the federal government.  

While mentioned in the CDAO press release, OpenAI has had its own $200 million DoD frontier AI deal since last month, when the DoD announced a pilot program to see the ChatGPT maker develop unspecified frontier AI products for the Pentagon.

OpenAI confirmed to The Register that the $200 million announcement it shared in June is the same award as its June one, making this more of an expansion to the earlier announcement than a new deal for the company. 

To sum it all up, it looks like employees can protest defense contracts all they want, and companies can insist they don't want to be involved in defense work, but at the end of the day, it's pretty tough to resist Uncle Sam's money and all that sweet, sweet data. Welcome to the military industrial complex, folks. ®

Updated to add

The DoD has been in touch to answer some of our questions, namely about how it intends to use the capabilities of each company differently - it’s less a competition between leading-edge AI firms and more a way to see how each approaches a problem to find the most efficient solution.

"Each frontier AI company has different strengths to offer the Department," a Pentagon official told The Register in an email.

The official added that each company will spend a few weeks scoping out potentially AI-solvable problems across the military that they’ll then present to Department officials. Future work after that initial trial period will be awarded based on cost, feasibility, and mission impact, the DoD said.

"We expect each company will offer unique value to the Department and we will leverage each of their strengths," The spokesperson told us. "This work will help integrate trusted commercial AI capabilities directly into DoD decision-making environments and workflows."