The rise of the military-technology complex
How Silicon Valley tech firms remade US defense contracting. And then opened the Pentagon’s checkbook—wide.
Logos of major tech companies with defense contracts (left to right: Meta, xAI, Google, Microsoft, Anthropic, OpenAI) added to a photo of Pentagon plaques showcasing the US armed services. Photo illustration by Thomas Gaulkin; source photo by Celal Güneş / Anadolu
The tale of how the Pentagon lost its spirit of innovation often begins with an incident known as “the Last Supper.”
It was the fall of 1993. About two dozen executives from leading defense contractors had received an unexpected invitation to dinner at the Pentagon. After the meal, the group was ushered into a briefing room where William Perry, then the deputy secretary of defense (and soon, the defense secretary), gave a presentation that startled the room.
In the wake of the Cold War, he said, the defense industry should prepare for catastrophic cuts to military spending. “They should not look at it as a little dip here but as a new low plateau and adjust their plans accordingly,” Perry would later recall.
Within a couple of decades, the US defense industrial base shrank from dozens of firms to today’s so-called Big Five—Lockheed Martin, RTX (formerly Raytheon), General Dynamics, Boeing, and Northrop Grumman, which together receive about a third of the Defense Department’s annual contract obligations.
Two decades of defense contractor mergers and acquisitions consolidated the defense industry from more than 75 firms to the Big Five—Lockheed Martin, Boeing, Northrop Grumman, General Dynamics, and Raytheon (now RTX)—also known as “primes.” Chart by Thomas Gaulkin. Source: NASA, “Final Report of the Commission on the Future of the United States Aerospace Industry”
Silicon Valley executives, like Shyam Sankar of the tech firm Palantir, have pointed to that dinner as the start of a decades-long “decoupling of commercial innovation from defense.” The chill between the Defense Department and Silicon Valley grew so strong that Sankar dubbed it “the Great Schism” in a 4,000-word, viral-among-some manifesto he called The Defense Reformation.
During those years of estrangement, the Pentagon became increasingly entrenched in the status quo, buying and maintaining hulking weapons within a ponderous bureaucracy. Silicon Valley, for its part, dashed ahead with innovations around cybersecurity, artificial intelligence, and product design—though usually to enhance consumer experiences in shopping, gaming, or food delivery.
The distrust and disdain between the two camps grew serious. “There’s plenty of people in Washington in the last 20 years who’ve made a point of how much, you know, they hate us,” Netscape founder-turned-VC-titan Mark Andreessen said in a 2025 podcast.
By the early aughts, there were people in both the Pentagon and Silicon Valley who knew that mending the breach was urgently needed. China’s leaps in technological advancement sounded an alarm in foreign policy circles. In October 2024, Sankar, who is Palantir’s chief technology officer, was still making a case for the reunification of tech and defense when he published the Defense Reformation.
Sankar began the essay by summoning the story of the Last Supper as a way to explain why the country was in a “state of undeclared emergency.” Consolidation had led to the demise of the defense industrial base, he argued. The China threat meant robust innovation adoption at the Pentagon was an imperative. To get there, speed and deregulation were required.
“We need a defense Reformation to … transform the way the government does business,” Sankar wrote.
Shyam Sankar, then Palantir’s Chief Operating Officer, gives a TED talk on “the rise of human-computer cooperation,” June 26, 2012. James Duncan Davidson / TED
Sankar and Palantir, the controversial data-mining company that provides AI-enhanced platforms for battlefield analysis, are powerful voices in a growing chorus. Anduril, which makes autonomous weapons systems, issued a 7,000-word decree pushing for policy changes, such as aligning Pentagon tech purchasing with “industry best practices” to “reboot the arsenal of democracy.” Venture capitalists like Andreessen Horowitz are investing upwards of $1 billion in defense tech, demanding deregulation because “America’s most innovative founders are building the tools our warfighters need—but the current system locks them out.”
In the past year, there have been indications that the Silicon Valley crowd is being heard. “Industry has become influential in shaping national security policy, with leading AI companies framing too much regulation as a national security liability,” said Emile Ayoub, senior counsel with the Brennan Center’s Liberty and National Security Program.
Within months of entering office, President Donald Trump issued a flurry of executive orders to accelerate AI adoption and streamline defense buying. To transform the US military into “an ‘AI-first’ warfighting force,” Defense Secretary Pete Hegseth has mandated faster but riskier methods of acquiring frontier technologies. “This is the beginning of an unrelenting onslaught to change the way we do business and to change the way the bureaucracy responds,” Hegseth has said.
After giving a speech at the Special Operation Forces Week convention on May 6, 2025, US Defense Secretary Pete Hegseth visits with Palantir executives at the company’s booth. Hegseth has called for more risk-taking in both the government and private sector to make the military an “AI-first" force. @PLTRs_Palantir / X
For example, the Pentagon has prioritized the use of Other Transaction Authority agreements, which cut red tape but are also the perfect way to facilitate sweetheart deals, one expert told the Bulletin.
It’s too early to see how those policy changes will translate on a broad scale. But nine notable AI firms—each chosen because the company received recent awards from the Pentagon’s Chief Digital and Artificial Intelligence Office or owns a sizable stake in a company receiving the award—have jointly had more than $5.8 billion earmarked for their work with the Defense Department in the last fiscal year (the contract ceilings are much higher). These companies represent a small subset of defense tech firms doing business with the Pentagon, and they account for a relatively small fraction of the Pentagon’s nearly $1 trillion budget. Still, their work with the Pentagon hints at the remarkable expansion in the defense-tech sector.
But Silicon Valley’s growing business prospects with the Pentagon haven’t been based on just words and ideas. Money and relationships have also paved its return.
AI-related defense tech contractors have been generous in their financial contributions to elected officials. The support goes well beyond the most conspicuous displays, such as Big Tech’s eye-popping donations to Trump’s re-election campaign, inauguration, and ballroom.
In fact, in all of the traditional follow-the-money ways, AI defense tech contractors have notably increased their spending on political influence, according to a Bulletin analysis. Between 2022 and 2024, these nine trailblazing AI companies and their employees have increased their campaign contributions by about 60 percent to the members of Congress who control Pentagon policy. Their defense-specific lobbying has jumped from about $650,000 to $7.7 million in the past decade.
These numbers are modest compared to what the Big Five spends—we’re talking millions, not hundreds of millions—but it’s a sign that the sector is at an inflection point.
US Army Chief of Staff Gen. Randy A. George administers the Oath of Office to four tech executives: left to right, Andrew Bosworth (Meta), Bob McGrew (formerly OpenAI), Shyam Sankar (Palantir), Kevin Weil (formerly OpenAI, now Meta). They are assigned to the Army’s “Executive Innovation Corps” to drive tech transformation in the military. Staff Sgt. Daniel Hernandez / US Army
Simultaneously, the Trump administration has installed a revolving door between defense-tech companies and the Pentagon. The deputy secretary of defense, Steve Feinberg, is the cofounder and former CEO of a private equity firm that invests heavily in companies with Pentagon contracts. The President’s Council of Advisors on Science and Technology co-chair David Sacks is a venture capitalist who invests in defense tech. The country’s science and technology advisor, Michael Kratsios, was previously an executive at top AI defense firm, Scale AI. The armed services are also stacked with senior leaders with ties to the tech industry, such as Under Secretary of the Army Michael Obadal, an Anduril alum.
Silicon Valley’s growing leverage with the Pentagon has come into starker relief in the last year, but it belies the fact that changes over the past two decades—across Republican and Democratic administrations—have taken industry beyond mere reconciliation to an optimal version of Sankar’s reformation. Indeed, tech firms are now positioned to do deals faster, under less scrutiny, with a customer that wields a trillion-dollar-and-growing bank account.
Or as Michael Dodd, the assistant secretary of defense for critical technologies—who also happens to be a venture capital advisor—recently assured industry: If companies move quickly from “concept to combat,” “the Department’s checkbook is open.”
William Hewlett and David Packard survey construction of the Hewlett-Packard facilities at Stanford’s Industrial Park (now called Stanford Research Park) in 1957. The company founders were both students of Fred Terman, a Stanford engineering professor who conceived of the industrial park in the wake of World War II and is considered one of the founding fathers of Silicon Valley. Hewlett-Packard Company Archives
A view of the Santa Clara Valley with prune orchards blooming near San Jose, from a 1927 travel brochure by the Southern Pacific Company.
Once a landscape of orchards and cattle ranches, the region now known as Silicon Valley was transformed by defense funding. The area would get its nickname from companies like San Jose’s Fairchild Semiconductor, founded in 1957, which received defense contracts to make silicon semiconductors.
The very foundations of computing and the internet were developed via funding from and collaboration with the Defense Advanced Research Projects Agency. More recently, DARPA support led to self-driving cars, Google, and iPhone’s Siri.
But sometime during the 1990s, starting soon after the Last Supper, Silicon Valley and the Defense Department were increasingly on the outs. Traditional defense firms like Lockheed Martin were still set up in the Valley, reliably pulling in federal funds. The next generation of tech companies, though, was focused on the expanding global consumer market, developing short-lived companies like Pets.com or the ice cream-and-video-game delivery service Kozmo. These consumer-tech companies fed the dot-com boom of the 1990s, that fleeting period of frothy startups and extravagant Bay Area launch parties.
By the time the bubble burst in the early 2000s, much of the Valley had entered a period of denial about its history with the Defense Department. Its foundation and roots had come from defense R&D dollars, but even after the dot-com crash, the new founders of social networks and cloud computing were making too much money innovating for consumers and B2B clients to seek work from the Pentagon. It wasn’t so much that Silicon Valley and Washington weren’t talking; it’s that the tech industry had forgotten—or perhaps was content to ignore—its past dealings with the Pentagon. As historian Margaret O’Mara puts it, “[i]n the sun-dappled quads of Stanford and the polished-wood wine bars of University Avenue, the only history that seemed to matter was that of Sergey and Larry and Mark or perhaps, if you wanted to get nostalgic, the two Steves in their Los Altos garage.”
There were those from the Valley, often incubated at Stanford University, who have long disagreed with this direction. In particular, Peter Thiel and Alex Karp, two of the founders of Palantir, have disparaged their peers for focusing too much on cat videos and not enough on defending democracy.
This is a foundational point in Karp’s 2025 book, The Technological Republic: Hard Power, Soft Belief, and the Future of the West. “A generation of founders cloaked themselves in the rhetoric of lofty and ambitious purpose … merely to build photo-sharing apps and chat interfaces for the modern consumer,” he writes.
Palantir CEO Alex Karp presides over a November 3, 2025 company earnings call. YouTube / Palantir
Karp ran decisively in the other direction. In 2003, he helped launch Palantir, which has been direct about its pro-defense posture, selling software that helps the intelligence community and the Pentagon parse the massive amounts of data the government collects. Palantir got a boost when it received about $2 million in an initial round of funding from In-Q-Tel, the intelligence community’s strategic investment arm, and by 2008, it began doing business with the national security branches of the government. It touts the fact that in 2020, it became the first defense tech company to go public in the 21st century.
In the past decade, Palantir’s revenue from defense contracts grew by an annual average of 50 percent, among the fastest of any defense firm, according to a Brennan Center analysis. That translates to at least $1.4 billion in defense department payouts since 2008.
Thanks to a surge in work with the military, Palantir became profitable in 2022. And its slice of the Pentagon pie continues to expand. Recently, the Defense Department embedded Palantir’s AI platform, used for battlefield awareness and target recognition, throughout the military. Now, as a “program of record,” Palantir will receive ongoing and long-term funding from the department.
Screenshots of Palantir’s “Maven Smart System” presented by Cameron Stanley, the Defense Departments’ Chief Digital and Artificial Intelligence Officer, at the company’s AIPCon showcase on March 12, 2026. Stanley said the system is now deployed across the department. YouTube / Palantir
At the time of Palantir’s founding, its focus on the national project was an anomaly. As Karp has observed, “A skepticism of government work and national ambition took hold in the Valley.”
The Bay Area had, indeed, become wary of the government, beginning with the Vietnam War. The conflict in Iraq solidified that position. Not inconsequentially, the 2013 Edward Snowden revelations—that the federal government had hacked tech companies to spy on its citizens—pitted Silicon Valley against the government over privacy issues. A few years later, Apple cited privacy concerns when it refused the FBI’s request to unlock the iPhone of a San Bernardino gunman who killed 14 people.
Christian Brose, a former staff director of the Senate Armed Services Committee, has written that in Washington, this refusal was viewed as yet more evidence that Silicon Valley was “morally unserious and willing to elevate corporate profits above national defense” even as its tech firms were willing to do business with China.
Alliances were nevertheless forged between some portions of Silicon Valley and the federal government. The long and sometimes forced relationship between the two is chronicled in historian Margaret O’Mara’s book, The Code: Silicon Valley and the Remaking of America.
As the Valley grew into a hub of global capitalism, O’Mara observes, Washington looked to the tech industry as an economic engine and a source of campaign financing. It courted Silicon Valley dollars by offering tax breaks, subsidies, and skilled worker visas and promised to fill classrooms with computers and internet connections.
No longer a group of mere hobbyist tinkerers, Silicon Valley needed to defend its business interests against competition from Japan, tax increases, and regulation of the internet. Valley companies hired lobbyists, gave sizable campaign contributions, and made friends with legislators from Newt Gingrich to Al Gore. Even tech leaders who had once prided themselves on being apolitical—Steve Jobs once famously bragged that he’d never voted—eventually went on to make the rounds in Congress.
As Silicon Valley gained prominence in the global economy, its producers and investors increasingly allied with government policymakers, like Newt Gingrich (seen on the August 1995 cover of WIRED) and Al Gore (on the May 2006 cover).
It would be wrong, of course, to say that the last 30 years was a period of complete technological stagnation at the Pentagon.
Instead, efforts at innovation have been a “slow build,” said Erin Dumbacher, a senior fellow at the Council on Foreign Relations. Dumbacher noted that many defense leaders over many administrations “understood that there was a gap in what military capabilities offered versus what commercial could do.” From the controversial Project Maven, which began integrating AI into battlefield systems, to the Replicator initiative, which sought to scale new capabilities such as low-cost drones, the Pentagon undertook myriad efforts aimed at modernizing the military.
When Donald Rumsfeld served as secretary of defense under George W. Bush, he went so far as to declare war on bureaucracy to bring innovation back to the Pentagon. “Our process and regulations have become so burdensome that many businesses have simply chosen not to do business with the Department of Defense,” he said. “To transform the Department, we must take advantage of the private sector’s expertise.”
Rumsfeld delivered that speech on September 10, 2001. The next day, the Twin Towers fell, and American Airlines Flight 77 crashed into the west side of the Pentagon.
The terrorist attacks of 9/11 and the subsequent focus on the Global War on Terrorism meant attention and dollars went to developing technologies to solve specific and immediate problems. The Defense Department quickly developed the Mine Resistant Ambush Protected Vehicle to protect soldiers from improvised explosive devices in Afghanistan and Iraq. Improvements were also made to uncrewed aircraft like the MQ-9 Reaper, the first so-called hunter-killer autonomous aircraft, which could gather intelligence and deliver munitions. It was a step toward today’s drone warfare.
A “retro advertisement” for the MQ-9 Reaper drone at Holloman Air Force Base, New Mexico. Antonio Salfran / US Air Force
Dumbacher, who has also served as an advisor in the Pentagon’s Office of the Under Secretary for Policy and as a director at In-Q-Tel, said that the gaps in the Pentagon’s tech uptake have much to do with the government’s prioritization of conflicts in the Middle East.
“Huge percentages of the defense budget were going to the wars in Iraq and Afghanistan, and the technology that we needed for those wars,” she said. “We are now 30 years from not having invested back in that strategic area, and it’s not just because we were living in a peace-dividend world. It’s really because we were focused elsewhere.”
Still, the Pentagon has long been aware that it needed smaller, lighter, automated weapons that were networked, enabled by AI-enhanced software, operational through the cloud, and secure from cyberattacks. This is now eminently obvious, given the way war has been waged in Ukraine, Gaza, and Iran.
The US military exists within a strange contradiction. It has some of the most formidable weapons on the planet and has pioneered innovations like artificial intelligence and autonomy. At the same time, it often lacks the infrastructure and process to deploy leading-edge technology quickly and at scale.
According to Brose, who was also a senior policy advisor to Sen. John McCain and is now the chief strategy officer at Anduril, too many of the previous attempts aimed at modernizing the military have focused on producing “better versions of the same kinds of platforms that the US military had relied upon for decades.”
“I have watched individual service members juggling a dozen separate chat windows, which can often involve taking information generated by one machine and manually transferring it to another machine,” Brose wrote in a 2020 book about the military's need to adopt “high-tech warfare” to guard against conflict with China. “They call it ‘hand jamming’ and ‘fat fingering.’ It is slow and prone to human error.”
Brose’s book is called The Kill Chain, the indelicate military term used to describe the decision-making loop that ends with an adversary's demise. Recent advancements in AI have revolutionized the tempo of the kill chain—information needs to be synthesized far faster than hand jamming can accomplish.
This is the new reality, says Frank Kendall, who spent decades at the Pentagon under both Democratic and Republican administrations. “When technology provides an advantage in warfare, it's adopted,” said Kendall, whose book, Lethal Autonomy: The Future of Warfare Whether We Like It or Not, was published this month. “What we have to do is figure out how to manage it, and how to ensure that the laws of armed conflict are still applied.”
Christian Brose, Anduril’s chief strategy officer, speaking on Fox News in May about the company’s new missile contracts with the Defense Department. YouTube / Fox News
An Anduril promotional video posted on YouTube in the summer of 2026 pairs quotes from FDR, JFK, and Ronald Reagan with clips of the company’s next generation military technology. YouTube / Anduril
Other proponents argue that speeding up the kill chain isn’t just about killing. “[R]esponsible technological investment in the defense sector is critical to our ability to deter unnecessary wars,” according to Trae Stephens, a cofounder of Anduril and a partner at Founders Fund, the storied venture capital firm.
If that argument sounds familiar, it’s because it echoes the logic of building and maintaining a nuclear weapons program. Alex Karp calls this “our Oppenheimer moment.”
Pentagon and defense technologists have said that combat decisions will not be made by robots alone, and humans must remain accountable for AI-related decisions and outcomes. Eric Schmidt, former CEO of Google, helped develop the Pentagon’s AI ethics protocols in 2019 as the chair of the Defense Innovation Board.
Meanwhile, organizations ranging from the United Nations, to the Pope, to a consortium of nonprofits and academics known as Stop Killer Robots, decry the use of AI in warfare, especially in the deployment of lethal autonomous weapons. They argue that there’s an inherent lack of transparency when the black box of AI models is further obscured by the opaque or classified nature of military operations.
“The technologies we’re worried about reduce living people to data points,” according to a statement from Stop Killer Robots.
Critics of the military’s use of AI say they take no comfort in the anti-regulatory posture of the current White House, which sought to prevent state AI regulations from taking hold and did away with a Biden executive order aimed at its “safe, secure, and trustworthy” development. President Trump recently signed an executive order requesting that companies voluntarily give the government 30 days to review new AI models before they’re released to the public—after former White House AI czar and VC investor David Sacks made adjustments and then gave his blessing to the order, a softened version of what the Trump White House had previously floated.
Some of the most ardent pushback against AI warfare has come from within Silicon Valley itself. Indeed, despite Thiel and Karp’s admonitions, among rank-and-file engineers discomfort with defense work still exists.
Google workers hold signs during a global company walkout at Jackson Square Park in New York City, on Nov. 1, 2018. The walkout followed a New York Times report that former Google executives like Andy Rubin received multi-million dollar severance packages after being accused of sexual misconduct, adding to ongoing protests of Google’s work for the Pentagon—including the Project Maven AI system and JEDI cloud data programs featured on the sign at front. Peter Foley/Bloomberg via Getty Images
In 2018, thousands of Google employees protested the company’s work on Project Maven, the Pentagon’s first attempt to integrate AI and machine learning into defense intelligence and operations. Maven was first deployed to analyze video and images—an innovation called computer vision—to monitor ISIS.
The episode was a PR nightmare for Google, with letters and op-eds leading to headlines such as, “Google’s march to the business of war must be stopped.”
The whole affair threw cold water on what had been slowly warming relations between the Pentagon and Silicon Valley. It also revealed the fissures between tech executives, who saw opportunities in defense contracting, and their employees, who had embraced Google’s early “Don’t be evil” motto.
A former Google employee told me that the company’s turn toward military work had made her “despondent.” “Rather than even trying to tell better lies to us about how we were abiding by AI principles, they moved in the other direction,” Vanessa Moran said of the company, which eliminated a ban on using the technology for weapons in 2025. Moran quit last August. Google did not respond to a request for comment.
In the end, the protests and resignations didn’t stop Project Maven from moving forward. Google did not renew its contract, but Palantir, Anduril, Amazon Web Services, and Anthropic continued on with the work. It’s now a permanent program of the US military.
But the within-the-Valley activism against Pentagon encroachment continues, and Google employees remain on the frontlines. After the Anthropic-Pentagon dispute over how the AI company’s models could be used by the Defense Department, nearly 1,000 employees at Google, plus about a hundred more from OpenAI, signed an open letter to make their concerns clear.
“We hope our leaders will put aside their differences and stand together to continue to refuse the Department of War’s current demands for permission to use our models for domestic mass surveillance and autonomously killing people without human oversight,” the letter says. (OpenAI has said in a statement that its technology will not be used for domestic surveillance or autonomous weapons.)
Thousands of Google and OpenAI employees have signed an open letter opposing the use of their AI models for “domestic mass surveillance and autonomously killing people without human oversight." notdivided.org
OpenAI employees are backing a new political action committee, Guardrails Alliance, to advocate for more robust regulations on AI. The head of Google DeepMind has also recently argued in favor of a new regulatory agency that would oversee frontier AI development.
Nevertheless, Google and OpenAI—along with Nvidia, SpaceX and others—signed agreements with the Defense Department this spring to enable their AI capabilities on classified networks. Little detail has been made public about these agreements, but the work will “strengthen our warfighters’ ability to maintain decision superiority across all domains of warfare,” according to a statement from the Pentagon.
One of the companies joining this group of Silicon Valley heavy hitters is Reflection AI, a relatively new firm that receives VC funding from 1789 Capital. In yet another example of the red flags around conflicts of interest and self-dealing emerging from the Trump administration, Donald Trump, Jr., the president’s son, is a partner at 1789 Capital—in fact, the lead partner on the deal.
Former US Defense Secretary Ash Carter holds his doctoral thesis up for an audience at Stanford University, on April 23, 2015. In his speech he announced the launch of the Defense Innovation Unit Experimental (DIUx), which has become an important driver of innovation and integration with tech firms in Silicon Valley and beyond. Clydell Kinchen / Defense Department
One of the more overt signs that the scales of power were tilting toward Silicon Valley came in the spring of 2015, when the secretary of defense went to Palo Alto.
At a lectern in a wood-paneled lecture hall at Stanford University, Ash Carter, secretary of defense under the Obama administration, made an overt bid to win back the technologists. “We must renew the bonds of trust and rebuild the bridge between the Pentagon and Silicon Valley,” he said.
Carter had been calling for innovation in the military since the early 2000s, and his second-in-command, Deputy Secretary of Defense Robert Work, had created the intellectual scaffolding of the so-called Third Offset Strategy, which sought to pull innovations like AI, machine learning, and automation into military operations. The idea was that the United States needed to overmatch—or offset—China and Russia’s tech capabilities, and it forged the realization that the Pentagon was no longer the agent driving innovation. Indeed, the technology the department needed was being developed on the other side of the country.
Carter’s visit to Palo Alto followed on prior attempts to appeal to industry. There was Rumsfeld’s stated war against bureaucracy in the early 2000s. But going back to the ‘90s, President Bill Clinton made efforts to deregulate defense procurement and signed the Federal Acquisition Streamlining Act, which requires the Pentagon to seek out commercial products before building its own. (Palantir, though, frequently points out that in 2016, it had to invoke the act and sue the government to win a contract with the Army. The firm’s executives have declared that the act “may be the most violated law on the books.”)
But in that pivotal 2015 speech, Carter was making a blatant effort to meet Silicon Valley where it was and told the audience that he was launching the Defense Innovation Unit Experimental (DIUx). The unit would help facilitate deals between tech firms and the Pentagon, with an emphasis on getting contracts signed in months, instead of years. The accelerated timeline would help companies avoid the dreaded Valley of Death, or the lag time between a successful prototype and a government contract, which is when the effort becomes financially sustainable. (See sidebar.)
DIUx would set up shop in Mountain View, near the NASA research campus, less than four miles from the Googleplex. “They’ll strengthen existing relationships and build new ones; help scout for new technologies; and help function as a local interface for the department. Down the road, they could help startups find new work to do with DoD,” Carter said.
It was what Silicon Valley needed to hear. “The skeptics in the audience might not have been moved by the stuff about defending the Western liberal order—but they did care about money,” observed Raj Shah and Christopher Kirchhoff, two of the first leaders of DIUx, in their book Unit X: How the Pentagon and Silicon Valley Are Transforming the Future of War. “Carter’s message was loud and clear: The Pentagon was open for business and wanted to push a giant shopping cart down the aisles of Silicon Valley.”
To make those deals happen, DIUx needed to completely upend the way it bought technology. The Defense Department is a so-called monopsony, or the only buyer when it comes to missiles and aircraft carriers. It had become accustomed to calling the shots on how deals were structured and on production schedules. The tech companies weren’t having it.
“When you look at a place like Silicon Valley, not only is DoD not a single buyer but quite frankly, they’re not even an important minority buyer,” said Lauren Dailey, a former DIUx procurement specialist in a 2021 podcast. “We, as DoD, have to change the way that we do business to be able to attract them.”
As the story goes, Dailey was an Army acquisitions officer who’d been dispatched to the DIUx from Washington, and within a year, she’d figured out how to hack the bureaucracy. Dailey turned to a method of Pentagon purchasing known as “Other Transaction Authority” awards.
OTs are little-known outside of the wonky world of government acquisition, but they’re not a new invention. OTs were first used by NASA in the 1950s to help it compete with the Soviet Union during the space race. They’re a form of federal buying that allows the government to move fast precisely because standard contracting regulations need not apply. Neither does the Competition in Contracting Act nor the Contract Disputes Act.
Aside from eliminating red tape, there’s also great flexibility in how OTs are crafted. Everything, including intellectual property and data rights, is up for negotiation.
Traditionally, OTs have been used to fund R&D or prototypes, so in and of themselves, they couldn’t lead tech companies out of the Valley of Death. But Dailey had discovered something spectacular while reading the 500-plus-page 2016 National Defense Authorization Act. Congress had given the Pentagon the ability to move successful prototype OTs directly into production, which is the point at which the big and steady money flows in.
The text of an amendment to the 2016 National Defense Authorization Act is fewer than 150 words but it encouraged the growth in the use of Other Transactions Authority agreements at the Defense Department, which totals more than $100 billion since 2016.
In other words, companies on a prototype OT could now jump directly to a long-term deal with the Pentagon, avoiding slowdowns related to the ornate accounting used to show fair pricing and evading time-consuming bid protests from rejected competitors. Indeed, it’s currently unclear to what extent the Pentagon’s decisions on awarding an OT can be contested in court.
Valley of Death averted.
But the Pentagon still needed a way to get the companies to vie for the work. The unit started pairing OTs with something called the Commercial Solutions Opening, a Shark Tank-like solicitation process in which the Pentagon tells industry about the problem it seeks to solve, and tech companies pitch their solutions through white papers, slides, and video demos. They’re appealing because they allow an agreement to be signed quickly without a formal bid and proposal process.
A CSO also allows the defense department to evaluate proposals based on their merits rather than on a formal set of pre-determined criteria. Multiple entities can be selected through a single CSO. And since this process is viewed as competitive, it allows companies to go directly from a prototype OT to a production OT later on with minimal bureaucratic friction. And it avoids another round of competition.
The “x” in DIUx stood for “experimental” but about three years after it was formed, the “x” was dropped from the unit’s name, and it became a permanent function of the department. The DIU now operates with a nearly $1 billion budget.
By then, the unit had awarded $354 million to 104 contracts by relying heavily on OTs and CSOs. The use of OTs by the defense department has seen steady growth, from about 1,700 awards in 2019 to 7,400 agreements in 2024, which hit $18 billion that year.
Upping the use of OTs is often on the wish list when Silicon Valley engages with Congress or the defense department. In 2023 testimony before a Senate Armed Services sub-committee, Palantir’s Sankar said that the Defense Department should become more comfortable using OTs. The same year, the Software in Defense Coalition sent a letter to the Senate Armed Services Committee, to argue that the department had failed to rely on OTs “to their fullest extent.” Andreessen Horowitz promoted them in a 2025 blog post. And going back to 2018, Eric Schmidt praised DIUx’s use of OTs in testimony before the House Armed Services Committee, and then recommended their use as the chair of the Defense Innovation Board in 2019. (Schmidt also served as chair of the National Security Commission on Artificial Intelligence, where he pushed for increased defense spending on faster adoption of military AI and was criticized for simultaneously investing in AI startups through his venture capital firm, Innovation Endeavors.)
In the last decade, Defense Department contracts have increasingly been awarded to smaller companies under the department’s expanded Other Transactions authority. From fiscal year 2021 to 2024, Defense OTAs increased from nearly $12 billion (excluding Covid-related funding) to more than $18 billion, and award for follow-on production OTAs nearly doubled from $1.2 to $2.2 billion. Chart by Thomas Gaulkin / Flourish. Source: GAO, “Other Transaction Agreements: Improved Contracting Data Would Help DOD Assess Effectiveness.“
While they have benefits, OTs also come with risk. Aside from the wide latitude allowed in drafting the agreements, there’s less oversight to guard against waste, fraud, and abuse.
When Frank Kendall was the Under Secretary of Defense for Acquisition, Technology and Logistics during the Obama administration, he encouraged the use of OTs in some cases. But he also said they’re the perfect way to facilitate sweetheart deals. There’s “a potential for abuse,” Kendall said. “There’s risk of corruption. There’s risk of a poorly written contract that makes it very hard to hold the contractor accountable for their performance.”
There’s also a question as to how and whether OTs are delivering for the warfighter and taxpayers. In 2021 and 2022, the Defense Department’s inspector general found that the Pentagon was not comprehensively tracking OTs and did not know how many there were, their dollar values, and which ones had moved from prototype to production. In September 2025, the GAO once again found that the Defense Department did not have a good handle on all of its OT data.
The public reporting on OTs is limited. A list of transactions is sent to Congress every year, but those reports are not made publicly available. They are also not accessible through USA Spending, the database commonly used to view government contracts.
Whatever the pros and cons of OTs, they are now a preferred method for buying software at the Pentagon, following a 2025 Hegseth directive.
The shift to OTs is notable, but it’s only one piece of a larger effort to dismantle some of the acquisitions bureaucracy that has irked Silicon Valley. Hegseth has also eliminated the Joint Capabilities Integration and Development System, an entity often blamed for the Valley of Death because it dictated fine-grained requirements before the department could buy and build. Congress also shifted the responsibility of Pentagon buyers from focusing on individual programs to overseeing a portfolio of related projects, so schedules and funds can more easily be shifted around.
Secretary Pete Hegseth holds a copy of the Defense Department’s “Acquisition Transformation Strategy" report during an address to military leaders and defense industry executives at Fort McNair, Washington, D.C., on Nov. 7, 2025. Alexander Kubitza / Defense Department
At a November 2025 speech at the National War College, Hegseth said these changes are meant to take a gamble. “We mean to increase acquisition risk in order to decrease operational risk,” he said.
It was the culmination of what Sankar and others had long asked for. Earlier this year Sankar co-published a book titled, Mobilize: How to Reboot the American Industrial Base and Stop World War III, that expands on the Defense Reformation. It was another call for speed, deregulation , and competition to maintain military superiority over China.
“The pieces for the Defense Reformation are in place: the capital class has shown up, the founders are busy building, and the Department of War is ready to innovate,” Sankar and a co-author wrote. “And not a moment too soon as we face the worst threat since the Cold War. Will we look back at this moment and realize that World War III had already started, and we just didn't know it?”
A cascade of tech company logos in a clip from an In-Q-Tel video celebrating its 25th anniversary in 2024. Conceived as a funnel for CIA investment in tech startups, the non-profit venture capital firm is currently the largest funder of private national security companies in the United States. YouTube / In-Q-Tel
To understand what’s driving the renewed relationship between Silicon Valley and the Pentagon, it’s instructive to look at where venture capital is flowing. Between 2024 and 2025, venture capital investments to defense technology nearlydoubled, from about $27 billion to $49 billion, according to PitchBook, a repository of data on capital markets.
The Silicon Valley Defense Group analyzes the prospects of the top VC and private equity-backed defense tech firms, and its 2026 report found that In-Q-Tel is still the largest funder of defense tech. But BlackRock, the world’s largest investment firm, significantly increased its focus on this sector in 2025. Not to be left out, the Big Five have also thrown funding at these companies through their own venture capital arms.
Overall, the Silicon Valley Defense Group observes “seismic” shifts in the sector. “[M]ore has changed in the past year than in any comparable period since the defense innovation movement began,” the report said. This follows on a decade of “dramatic growth for the emerging defense technology ecosystem.” The Pentagon is also actively seeking out input from defense tech investors through the newly created Defense Policy Board, which includes Marc Andreessen, whose firm runs the American Dynamism fund, and defense-tech investor and Thiel associate Blake Masters. (Palantir’s Theo Wold is also among the 15 members appointed this summer.)
Some startups are responding to the clear signals coming from the Pentagon. Sion Power, an EV battery company, recently announced a pivot to delivering products for military drones. Debut, a biotech company focused on the cosmetics space, landed a $2 million OT in 2024 to help the Pentagon develop a bioindustrial manufacturing facility.
Over the last decade, private investment in the top 100 privately held defense firms (as ranked by the Silicon Valley Defense Group) has increased from $230 million in 2016 to $86 billion in 2025. Chart by Thomas Gaulkin / Flourish. Source: 2026 NatSec100 report, Silicon Valley Defense Group.
Changes to the main product page on Sion’s website, seen in March 2026 and then June 2026, reflect the company's pivot toward defense. sion.com / archive.org
Elke Schwarz, a professor at Queen Mary University of London who studies military AI, observed that Silicon Valley has successfully reshaped how the Pentagon does business—and to its benefit.
“With its focus on iteration, agility, risk, speed, and scale, the new defence culture performs both the logics of its new AI products and that of the VC companies that fund the startups,” she writes. “It is a holistic shaping of the environment towards the vectors that make Silicon Valley VC investors enormously rich.”
She adds: “[M]ove fast and break things is not a suitable motto for any conscientious military organization.”
San José State University professor Roberto González has been studying the rise of defense tech from the heart of Silicon Valley. He said there’s an analogy to be made with the California Gold Rush. “That’s just a part of the mentality of this part of the world—you have an opportunity to get rich quick, and you go for it,” said González, author of War Virtually:The Quest to Automate Conflict, Militarize Data, and Predict the Future.
But, he warns, there’s a crucial difference between the two.
“What we’re talking about, in the end, is taxpayer dollars,” González said of the money helping to fuel the defense tech boom. “A lot of the startups will fail. But in the meantime, the pot can just keep growing larger and larger, as long as we can be convinced, as taxpayers, to keep that Pentagon budget growing.”
Bernice Yeung is the managing editor of the Investigative Reporting Program at UC Berkeley Journalism. She was born and raised in the Silicon Valley.
Editor’s note: The author of this article, Bernice Yeung, has signed on to the settlement of a lawsuit against Anthropic AI, which agreed last year to pay $1.5 billion to settle a copyright infringement claim brought by a group of authors whose books were used as AI training data. Under the settlement, Anthropic will compensate authors around $3,000 for each of the estimated 500,000 books covered by the settlement. Yeung has not yet received a payment from the settlement.
Elizabeth Santos and Anjali Kumar contributed reporting.
Together, we make the world safer.
The Bulletin elevates expert voices above the noise. But as an independent nonprofit organization, our operations depend on the support of readers like you. Help us continue to deliver quality journalism that holds leaders accountable. Your support of our work at any level is important. In return, we promise our coverage will be understandable, influential, vigilant, solution-oriented, and fair-minded. Together we can make a difference.
Get alerts about this thread
0 Comments
Oldest
Recent Stories
