How AI is rewriting Silicon Valley’s relationship with the Pentagon

This article is an installment of Future Explored, a weekly guide to world-changing technology. You can get stories like this one straight to your inbox every Saturday morning by subscribing here.

It’s 2035. After decades of investment, the US military now has AI systems integrated throughout its operations. Autonomous drones, tanks, and fighter jets enhance US defenses, while leading (hopefully) to fewer civilian casualties and collateral damage. That’s one future, anyway.

AI warfare

In December 1941, Japan’s military attacked Pearl Harbor, bringing the US into World War II. Less than four years later, America ended the war by dropping two atomic bombs on Japan — a destructive demonstration of its technological supremacy over every other military in the world, at the time.

America has maintained its military supremacy ever since, through a combination of air power, economic might, and advanced weaponry, but the nature of war is rapidly changing, and it seems possible that the dominant military of the future won’t be the one with the most powerful weapons — it’ll be the one with the best AI.

To find out if the US can win this race, too, let’s look back at its AI history and the current trends that will decide whether it falls behind another world power.

Where we’ve been

Where we’re going (maybe)

The Manhattan Project is the quintessential example of America’s scientific prowess enhancing its military might. That path to the atomic bomb began, in part, with Albert Einstein sending President Franklin D. Roosevelt a letter warning that German physicists were developing their own bomb and urging him to make sure the US beat them to it.

If you’ve seen “Oppenheimer,” you know what came next: the military and some of the greatest scientific minds in the US and its allies joined forces for the Manhattan Project, leading to the development of the bomb that ended the war.

But what if all (or even most) of America’s top scientists had refused to participate in the Manhattan Project? Or if Roosevelt had ignored Einstein’s warning? How might the war have ended then? While the Allies were able to defeat Germany before the Nazi atomic bomb project succeeded, it’s likely that defeating Japan would have involved a massive invasion, with potentially millions of casualties.

A dilemma like this could still play out in the future, but with AI instead of atoms. 

We’re already seeing how AI can free civilians from dull, dirty, or dangerous jobs, and it could do the same for the armed forces, but the Department of Defense (DoD) is notoriously slow to adopt new technology — until 2019, US nuclear weapons systems literally operated on floppy disks.

This might not be an issue if all nations were leaving AI out of the military, but China — the US’s biggest military rival — is aggressively developing AI for military use.

“The Chinese Communist Party deeply understands the potential for AI to disrupt warfare, and is investing heavily to capitalize,” Alexandr Wang, the CEO of Scale AI, told members of Congress in 2023. “AI is China’s Apollo project.”

Not only has the US military been investing less in AI than China, but many of America’s tech leaders in Silicon Valley have little interest in developing AI for the DoD, some for moral reasons and others because small R&D contracts rarely lead to major production deals.

“We live in the first period in US history where our best, most innovative tech companies — with the largest pools of talent in our country — refuse to work with the military,” Palmer Luckey, founder and CEO of defense tech company Anduril Industries, told Bloomberg in 2023. “We’ve never been in a situation like that before, and people don’t understand how dangerous that is.”

“We’ve set a big goal … to field attritable, autonomous systems at a scale of multiple thousands.”

Kathleen Hicks

For better or worse, this standoffish relationship between the DoD and Silicon Valley is now changing.

Between 2022 and 2023, the military nearly tripled the amount of money it spent on AI, and in 2023, it announced Replicator, an initiative designed not only to get AI systems out of the lab and into the field en masse, but also establish the processes needed to ensure future tech innovations don’t succumb to the innovation valley of death.

“We’ve set a big goal for Replicator: to field attritable, autonomous systems at a scale of multiple thousands [and] in multiple domains within the next 18-to-24 months,” said Deputy Secretary of Defense Kathleen Hicks. 

“The ‘replication’ isn’t just about production,” she continued. “We also aim to replicate and inculcate how we will achieve that goal, so we can scale whatever is most efficient, effective, and relevant in the future, again and again.”

Silicon Valley, meanwhile, seems to be warming to the idea of working with the Pentagon — a growing number of companies are now focused on developing defense tech, with billions of dollars in backing from venture capitalists.

Anduril has been at the forefront of this trend.

Co-founder Trae Stephens told Freethink he and his partners created the company in 2017 in response to the reluctance of leading Silicon Valley firms to work with the DoD “despite having the necessary talent, funding, technology, and infrastructure.”

In the years since, Anduril has landed multiple DoD contracts, including ones to develop autonomous underwater vehicles for the Navy and autonomous drones that could fly alongside piloted aircraft for the Air Force. It also secured an agreement worth up to $1 billion to develop, integrate, and maintain counter-drone systems for America’s special operations forces.

As for why other tech startups are warming to the idea of developing AI for the military now, Stephens says the war in Ukraine likely served as something of a wake-up call.

“Russia’s invasion of Ukraine disproved the prevailing globalist notion that economic ties alone could maintain peace and prosperity, sparking a renewed interest in the importance of defense in Silicon Valley, particularly given the high tech nature of Ukraine’s resistance to Russia’s aggression,” he told Freethink.

The evolving attitude in Silicon Valley means the DoD now has access to a larger pool of tech talent, but whether it’ll be able to translate that into deployed AI systems remains to be seen.

If small contracts to develop new AI products don’t start leading to large production orders for startups, VCs could stop seeing value in funding defense tech startups. The pool of talent could then shrink back down as Silicon Valley turns its back on the DoD once again.

Between that potentiality and the scale and speed at which China is investing in AI, the pressure is on for the Pentagon to not only get AI development right, but get it right right now.

“In the years ahead, American troops will operate under the shadow of sophisticated drones and networked sensors that can detect their movement and relay targeting data to precision weapons almost instantaneously,” Stephens told Freethink. 

“Now is the time for the US to both rapidly build and rapidly deploy AI technologies because our adversaries are not slowing down,” he added.

We’d love to hear from you! If you have a comment about this article or if you have a tip for a future Freethink story, please email us at tips@freethink.com.