Silicon Valley AI companies have a new BFF: the US Department of Defense.
Major companies developing creative AI technology have in recent months begun, deepened, or in some cases initiated relationships with the military revising internal policies or making exceptions to remove road blocks and restrictions for defense works.
Various DoD agencies, from the Air Force to various Intelligence groups, are actively testing use cases for Meta’s AI models and tools, googleOpenAI, along with technology from startups such as Anthropic and Mistral, Gladstone AI and ScaleAI, several people familiar with the trials said luck.
This is a significant development for Internet companies, which until recently treated defense work as taboo, if not outright. in verbs. But the cost of developing and running AI-powered services already runs into the hundreds of millions of dollars, and with no signs of slowing down, AI companies are feeling pressure to show some returns on their massive investments. The DoD, with its essentially unlimited budget and longstanding interest in cutting-edge technology, suddenly doesn’t look so bad.
While securing a contract with the defense can be difficult, with the layers of certification and strict compliance standards to meet, “the rewards are significant” and the money can roll in for years, Erica Brescia, managing partner at Redpoint Ventures, points out. AI investment, he said
“DoD contracts deliver significant annual contract values, or ACVs, and create long-term opportunities for growth and defense in the market,” Brescia said.
Brescia added that pursuing a Defense job has recently become more acceptable in tech social circles. Not only are business leaders eyeing the hundreds of millions of dollars in contracts being collected by defense-focused startups like Palantir and Anduril, but the “changing political landscape” has made “pursuing defense as the number one market segment an increasingly attractive option.” companies willing to navigate longer sales cycles and manage complex implementations.’
The military’s embrace of labor may come at a good political moment, with a pro-business Trump administration taking office in January and a group of Silicon Valley insider hawks, led by “First Buddy” Elon Musk, within the president-elect. the circle Musk’s tenure is in his official role as co-chair of the new Department of Government Efficiency to significantly reduce expenses. But few expect the Pentagon’s budget to face severe cuts, especially when it comes to AI. The US and China are locked in a battle for AI supremacy.
For now, much of the military’s work with generative AI appears to be small-scale projects and tests, but the possibility that generative AI technologies will become a fundamental part of computing in the future means that the relationship between Silicon Valley and the Pentagon could be huge. .
Defensive use of AI doesn’t have to mean drone warfare or blowing things up. Much of the AI-specific work within DoD is more mundane activity that any office would be happy to assign to a technology subject. Data tagging, collection, and sorting are common uses of AI in the department, as is the use of Chat-GPT and Claude chatbots, which most people can access online, but which require additional security when used by DoD. Large language models could also be useful for analyzing and searching for classified information, helping with government cybersecurity work, and providing better vision and autonomy for things like robotic tools, drones, and tanks.
Some tech companies try to avoid participating in DoD projects that could be used in the “kill chain,” a military term that refers to the structure of an attack on an enemy, a former DoD official told Fortune about the companies that won. recruitment contracts. However, sometimes concerns disappear as millions or billions of dollars become available. “Once you’re in, you want to expand,” the person added.
A changing set of ruless
Some technology companies, Palantir and on the sensor For years, Defense uses and contracts have become the focus of the entire business.
Within Silicon Valley’s established Internet companies and some of its younger AI startups, however, military work was shunned as companies sought to recruit and retain left-leaning engineering talent. When Google acquired it DeepMind In 2014, he reportedly made a commitment never use the startup’s technology for military purposes the goals And in 2018 he faced off against one Alphabet CEO Sundar Pichai internal reaction Over Google’s involvement in Project Maven, the Pentagon’s drone war effort. While Google insisted it was only using its technology for “non-offensive purposes,” such as analyzing drone footage, the outcry from employees was loud enough that Pichai canceled his vacation to pacify employees and finally order Google wouldn’t develop its own AI for weapons.
Chip Somodevilla/Getty Images
Google’s “AI Principles” it now specifies that it will “not seek … weapons or other technology whose primary purpose or implementation is to cause or directly facilitate injury to persons,” nor “surveillance that violates internationally accepted norms.” But politics leaves a lot to be desired, and the company has explicitly said it will not vow to work fully with the military.
The story is similar in other big AI players. The meta initially banned his Llama model from being used in military work, like OpenAI, while Anthropic initially built its Claude model “harmless“. Now, all three have announced that this work is fine with their models, and they are actively pursuing such uses. Sam Altman, who founded OpenAI the principle To develop AI “for the benefit of all mankind” and who he once said there were things he would “never do with the Department of Defense” ever since removed no commitment to restrictions arising from its usage policy.
A venture capitalist focused on investing in AI companies noted the “American Dynamism” of VC firm Andreessen Horowitz. essay two years ago was the moment when defensive recruiting avoidance began to change. The essay explicitly said that technology companies working in defense were working in America’s national interest.
“Executives started thinking, ‘Oh, OK, defending America, working with the military is really good,'” said the VC.
Widespread post-pandemic layoffs have also had a dramatic effect on worker protests, giving tech companies more freedom to conduct military business.
The DoD has already paid about $1 billion Official contracts to AI companies in the last two years, a luck the analysis While the details of the contracts are vague, they have been awarded to companies such as Morsecorp, which focuses on autonomous vehicle technology, and a subsidiary. ASGNa management and consulting company to develop new AI prototypes.
Not all such contracts are made public. But any government contract awarded to a major AI company would likely mean tens of millions or hundreds of millions of dollars in revenue for those companies and their biggest backers.
It is OpenAI’s largest investor Microsofthe recently said it was his Azure cloud service accepted DoD agencies to use OpenAI’s AI models for intelligence at lower security levels – something that took over. years of investment in specialized infrastructures. He is also the biggest sponsor of Anthropic Amazon. Amazon Web Services is perhaps the DoD’s largest cloud provider and has tens of billions of dollars in government contracts. For both companies, being able to add new AI services and tools to DoD offerings could be valuable. The same goes for a company like Google, which has also won valuable government contracts, and its Gemini AI model.
“They’re basically building the plane while they’re flying, so it’s a huge ground-breaker,” one AI executive told Fortune, eager for more tech companies to suddenly have their AI tools and models in DoD’s hands.
“Critical” technology for the DoD.
DoD defines AI is among its 14 “critical technology areas” because it has “tremendous promise” and is “essential to mastering future conflicts.”
About a year ago, the DoD officially created the Office of Strategic Capital, a new federal loan program in partnership with the Small Business Administration to ensure that critical technologies like AI receive funding through direct loans. For fiscal year 2024, OSC made $984 million available intention splitting into 10 companies focused on things like autonomous robotics and microelectronics manufacturing, which typically includes AI chip manufacturing. Another DoD is investing approx $700 million in chip manufacturing and creating domestic semiconductor manufacturing, which is key to creating AI chips.
Despite billions in investment and no signs of slowing down within Defense, the AI executive admitted that most of today’s AI products are “not that useful yet,” either for Defense or the general public. But applications at scale in a government or defense environment can be more useful, faster. “The military also effectively created the Internet.” ARPANET, the key technological foundation of the modern Internet, was built within the DoD, as were common technologies such as radar and GPS systems today.
Although a department like defense wants useful products, it also sees its budget as prestigious increasing year by yearHitting below $1 trillion in 2024 half of that budget it is assigned to companies that contract with the department.
“Honestly, yes, they really love throwing money around,” the executive said.
Additional reporting by Jeremy Kahn and Sharon Goldman.
Are you a tech company employee or someone with an idea or tip to share? Contact Kali Hays securely the signal at +1-949-280-0267 or at kali.hays@fortune.com.
