What Silicon Valley Gets Wrong About National Security
Margaret Mullins
MARGARET MULLINS is Director of Public Options and Governance at the Vanderbilt Policy Accelerator. She served as Senior Adviser to the Deputy Secretary of Defense in the Biden administration from 2023 to 2025.
In the summer of 1993, U.S. President Bill Clinton’s secretary of defense, Les Aspin, and William Perry, then the deputy secretary of defense, hosted a dinner at the Pentagon for defense industry leaders. The Cold War was over, they informed the gathering, and the federal budget would not support them all. With no looming Soviet threat justifying ever-rising defense budgets, consolidation would be necessary.
Just after it happened, this meeting was dubbed the “Last Supper” by the CEO of the Martin Marietta Corporation, a major aerospace and defense firm that went on to merge with the Lockheed Corporation in 1995. Other similar mergers would follow. The gathering has since become a convenient origin story for why the U.S. defense industrial base lost the ability to meet the military’s needs and the defense acquisition process became so onerous. According to this narrative, the Defense Department simultaneously meddled with and neglected the industry, creating the inflexible behemoths that dominate the sector today while cutting off pathways for emerging technology companies to compete.
Defense acquisition is, indeed, broken. The United States cannot produce critical materiel at speed and at scale in a moment of crisis. Despite spending more on defense than the next nine countries combined, the United States faces a crisis of both modernization and production. Recently, an emergent group of Silicon Valley defense tech leaders and their funders, including Alex Karp, the CEO of Palantir; Palmer Luckey, the founder of Anduril; and Katherine Boyle, a co-founder of the Andreessen Horowitz venture capital firm American Dynamism, have blamed excessive government regulation and interventionism for this unhappy state of affairs. Too much funding, they say, flows to large legacy programs and prime contractors. Some have put a spotlight on the Last Supper, arguing that it marked a turning point after which procurement became too slow, requirements too numerous, and risk-averse contract structures too stifling to innovation.
Their declinist narrative is convenient, but it is inaccurate. The Last Supper did not usher in the industry’s dark era. By the time Aspin and Perry held the dinner, the defense industry, like nearly all U.S. manufacturing sectors, had already been weakened by globalization and the financialization of the U.S. economy, drawdowns in the defense budget since 1986, and a broader effort to remake the government in the image of private business. Indeed, after the Last Supper, policymakers repeatedly did exactly what Silicon Valley’s tech leaders suggest today as a solution for the department’s ills: they deregulated the industry and outsourced production capacity to the private sector. In fact, such strategies helped create the very problems that now plague the industry.
Subscribe to Foreign Affairs This Week
Our editors’ top picks, delivered free to your inbox every Friday.
* Note that when you provide your email address, the Foreign Affairs Privacy Policy and Terms of Use will apply to your newsletter subscription.
Defense tech leaders’ alternative vision for procurement is likely to fail, just as previous deregulatory efforts did. Washington should not rush to accept Silicon Valley’s critique as gospel. Instead, it should accept that national defense is not a normal competitive market and never will be and invest in the government’s own capacity to oversee military production, incorporate new technology, and manage competition.
A FRAGILE JUGGERNAUT
The modern U.S. defense industrial base took shape following World War II, benefiting from the enormous investments made by the government and the decimation of foreign production capacity wreaked by the war. Before the war, the government had researched, developed, and produced most weapons in-house, at public arsenals, shipyards, and laboratories. During World War II, the government had enlisted the help of blue-chip companies such as Chrysler and IBM to manufacture a variety of defense-related products, from tanks to engines to rifles. This generally fruitful partnership between the military and industry endured after the war ended, undergirding the early Cold War effort to beat the Soviet Union in a technological competition and sparking innovation in national security with positive spillover benefits to the broader U.S. economy. Government investment and direction, paired with industry expertise, led to the creation and commercialization of inventions such as GPS, the Internet, and semiconductors.
By the 1970s, however, overseas manufacturing and technology markets began to reemerge. Foreign governments subsidized the domestic production of electronics, ships, aircraft parts, and more, and U.S. defense companies were forced to find ways to remain competitive. Taking advantage of lower labor costs overseas, U.S. producers moved parts of their supply chains to Asia and Europe. Then, the 1980s brought “financial engineering” to defense. Threats of hostile takeovers, defensive restructuring, and ruthlessly profit-maximizing executives saddled companies such as Lockheed and Martin Marietta with debt, leaving them weak and increasingly dependent on the Defense Department’s largess to stay afloat.
As a result, long before the Last Supper, the U.S. defense industry began to constrict. A defense spending boom at the heart of President Ronald Reagan’s hawkish Cold War strategy in the early 1980s masked the damaging effects that globalization and financialization had on the industry, but only temporarily. When Reagan began to draw down defense spending in 1986, the move only hastened the hollowing out of the bloated sector. A 1989 Center for Strategic and International Studies report found that between 1982 and 1987, the number of suppliers in defense subsectors fell by 67 percent.
The United States cannot produce critical materiel at speed and at scale.
In several key subsectors, the situation was particularly dire. The top four firms in the submarine business and the space booster business had 99 percent and 97 percent of the market share in their respective subsectors. The top four manufacturers of surveillance and detection satellites had 100 percent of their subsector’s market share. Meanwhile, the blue-chip conglomerates, including Chrysler, Ford, and IBM, were getting out of the defense industry. By 1990, all three had largely left the sector, judging that fewer contracts and declining budgets would make defense production less profitable than making automobiles and personal computers.
Instead of compensating for U.S. defense firms’ offshoring, consolidation, and constriction by bolstering domestic capacity, the government mimicked the private sector’s trends. For the Defense Department, this meant shedding its organic production capacity. Shifting the production of several core military capabilities to the private sector resulted not in a cost-efficient transfer of production but in the outright disappearance of some production capability. Take shipbuilding. Before World War II, public shipyards managed almost half the research, design, construction, and maintenance of the U.S. Navy’s ships. From 1953 to 1960, private shipyards’ share in new ship construction and repair contracts rose from 55 to 85 percent. But by the 1970s, some large private builders had stopped bidding on navy work. Instead of reinvesting in a public option, the Reagan administration further limited the government’s role, eliminating the construction subsidy program that had underpinned the commercial shipbuilding business since the 1930s. Domestic shipbuilding, for the navy or the commercial sector, never recovered.
Washington also privatized services and reduced personnel. Between 1981 and 1987, the Defense Department outsourced almost 40,000 service positions on military installations to private contractors, citing cost-saving concerns. But the contracts tended to be poorly structured and too narrowly defined, and the firms that took over the work often found that the actual cost of implementation was higher than anticipated. Because the contracts held the government responsible for additional costs, Washington frequently ended up paying more to these private firms than it would have paid its own civilian personnel. And by the time the contracts ended, the government, having shed in-house talent, could not revert to in-house provisioning.
SLASH AND BURN
In 1989, Jacques Gansler, a defense acquisition scholar and future Defense Department undersecretary, wrote that defense firms and the defense divisions of diversified companies suffered from “heavy debt, difficulty of borrowing, considerable excess capacity, low cash generation, high (and growing) risks, old production equipment, too little capital investment, relatively low productivity, mixed quality, and rapidly rising prices.” During George H. W. Bush’s presidency, the Defense Department warned that industrywide restructuring was inevitable.
At the same time, by the early 1990s, it was clear that emerging technologies increasingly developed in the private sector should play an important role in national security. The Defense Department would need a new approach to acquiring and incorporating this technology into the military. But rather than build up the Pentagon’s capabilities and invest in in-house talent and research and production capacity, the Clinton administration adopted a policy of “civil-military integration,” to be achieved by deregulating the industry, developing and encouraging the production of dual-use technologies applicable to both military and civilian purposes to create private markets for defense products, and procuring, whenever possible, commercial products rather than items created specifically for the military. With bipartisan congressional support, Clinton signed into law the Federal Acquisition Streamlining Act in 1994 and the Clinger-Cohen Act in 1996, which cut military-specific requirements and regulations, established a default preference for commercial technology, and simplified the acquisition of information technology.
Many of these reforms, however, did not have the hoped-for effect. Privatization, force reduction, and deregulation meant that ever-more complicated contracts had to be administered by fewer people and with less guidance. The outsourcing of information technology and data processing positions, cornerstones of the emerging private sector revolution, left the department less capable of understanding and adopting new technology. Reductions in force, which continued under a congressional directive in 1996, hit acquisition personnel especially hard. All told, between 1989 and 1999, the government’s defense acquisition workforce shrank by 50 percent.
Subsequent administrations failed to arrest the relative decline of government capacity. Between 2001 and 2015, the defense acquisition workforce increased by 21 percent while defense contract spending commitments increased by 43 percent. Between 2015 and 2020, the workforce grew by just 18 percent, as obligations grew by 41 percent. And although the Obama administration stood up the Defense Innovation Unit, the first Trump administration expanded use of other transaction agreements, and the Biden administration developed programs such as the Rapid Defense Experimentation Reserve and Competitive Advantage Pathfinders to improve the integration of emerging technology, none made any major reforms to state capacity.
CALL TO ARMS
Today, prominent voices in the new defense tech ecosystem and their funders argue that the Clinton-era deregulation simply did not go far enough. They see the government as fundamentally risk averse, beholden to taxpayers, and hostile to new industry entrants. Palantir’s chief technology officer, Shyam Sankar, for example, holds the Last Supper responsible for “the decoupling of commercial innovation from defense.” From the vantage point of these relative newcomers to the industry, the Defense Department’s byzantine processes, contract structures, and data requirements are shutting the door on a new generation of innovators. They argue that the Defense Department should limit military specification requirements for new technology, reduce compliance requirements for new entrants, and even increase competition within the department by having program offices compete with each other.
On the surface, their diagnosis is compelling. But these leaders ignore that their suite of preferred policies, which some winkingly refer to as a “First Breakfast,” have been tried already—and have failed. Many of the barrier-reducing measures would expose the Defense Department and American taxpayers to increased risk without an outweighing benefit. By easing regulatory and reporting burdens for new and commercial companies, the Defense Department would lose access to key data to determine whether a company’s pitch matches its product, whether that company can scale and deliver the product, and whether it is unfairly overcharging for the product. By using more flexible contracting structures for production, the Defense Department would be forced to rely on acquisition officers’ general understanding of companies, technology, and the market when they negotiate fair prices and terms, as opposed to direct insight into specific companies. And expanding the definition of nontraditional defense contractors may incentivize some more established legacy and nonlegacy companies to restructure superficially to get access to the department’s more flexible processes, meant to ease the entry of industry upstarts.
Defense tech leaders also argue that expanding the use of fixed-price contracts, which transfer cost risk (and upside) to the contractor, would allow firms to innovate faster. Unlike cost-plus contracts, which offer lower profit potential but allow the government to absorb the risk for systems requiring major technological advances or facing significant uncertainty, fixed-price contracts are typically used in deals for well-established and commercial technology with clear requirements. Historically, however, when the department has attempted to rely more on fixed-price contracts to save money and shift innovation risk to the private sector, costs have ballooned, development timelines have slipped, and the government has remained responsible for the risk.
Defense tech leaders have no reservations about pursuing consolidation.
Defense tech companies seem to consider themselves immune to the uncertainty that often compels the government to offer firms cost-plus contracts for emerging technology development because, they claim, their venture backers will absorb the risk, allowing them to achieve innovative leaps under fixed-price contracts. But over the long run, these firms and their funders are more likely to limit their own risk by shifting from defense-specific innovation (which invariably requires high, sustained R & D investment with uncertain outcomes that venture funds are unlikely to provide) to technology with clear commercial potential. Palantir, for example, evolved from an intelligence platform to support analysis for national security agencies to a data company that licenses access to its commercial software to government agencies and private companies. The growth of firms that offer competitive products to both the Defense Department and the commercial sector is, on the whole, positive. But such companies must ultimately answer to their profit-seeking investors and thus are unlikely to meet all of the department’s needs.
Furthermore, Silicon Valley’s proposed reforms would not guarantee increased competition. In fact, they could help large technology firms gain a foothold in emerging defense areas and then acquire new, smaller entrants. For all the invective against the industry consolidation that took place after the Last Supper, defense tech leaders have no reservations about pursuing consolidation themselves. Anduril, for example, has acquired nine companies since its founding in 2017. Palantir has acquired seven companies since 2013. AeroVironment, which manufactures small drones, has acquired six firms since 2019. Shield AI, which develops autonomous defense systems, has acquired three companies since 2021. Silicon Valley defense leaders argue that these moves are necessary to expand their offerings. In doing so, however, they echo the exact line that their more traditional corporate predecessors used in the 1980s and 1990s.
Finally, the solutions now being championed in Silicon Valley would not necessarily make the defense industrial base any more resilient in the face of a crisis. Often presented as a panacea for procurement modernization and production inefficiency, they would address only a narrow segment of the Defense Department’s needs. The capacity to produce less glamorous but strategically essential systems such as ships and munitions will not be solved with software alone, and they will require public investment in the government’s own expertise, production, and maintenance infrastructure—much of which has been deliberately dismantled over the past 40 years.
BACK IN BUSINESS
Defense tech leaders are correct that the U.S. military needs a faster procurement process and more agile development. But many of their proposed fixes would create a system that prioritizes speed and flexibility at the expense of fit, need, or cost discipline. Although few would object to a quicker acquisition process, the national defense requires viable, scalable, and surge-ready production, well-managed military-specific technology, and core capabilities resilient to market shifts in addition to speed. By advocating for “dynamism” as a cure-all, advocates distract from the more fundamental requirement to address the modernization and production crises: improving the structural capacity and expertise of the state.
To be sure, defense tech leaders do not totally ignore the need for state capacity entirely, but most treat it as one component among many, substitutable for private sector expertise, research, and production rather than the keystone of the entire system. To truly reform, Washington must defend itself against the false narrative that a lean government is an effective one.
If the Department of Defense wants to build a combined industrial base that can respond to the eruption of a conflict, it will have to pay for it. Washington should be willing to accept the higher costs that come with contracts that include provisions mandating that suppliers are capable of surging production if required. It should also ensure that the arsenals, shipyards, and depots that make up the Defense Department’s organic industrial base receive adequate funding. Critics may argue that expanding a near-trillion-dollar defense budget would be irresponsible, but it will cost less to maintain capacity in and outside government than to rapidly build or buy that capacity in a time of urgent need. Rather than relying on episodic crisis spending, Congress and the Defense Department should steadily invest in both the government’s and industry’s facilities, capital equipment, and manufacturing technology.
The government’s role is to advance the public interest, not to please its contractors.
The Pentagon also needs a process to assess when certain capabilities would be better developed in-house and when they would be better purchased from the private sector. Acquisition officers and others who must weigh these “build versus buy” decisions and negotiate increasingly complex contracts need more people, resources, and training to support them. To protect against waste and to encourage the responsible development and integration of new technologies in the private sector, the Defense Department should hire more acquisition officers tasked with conducting comprehensive market research and overseeing the adoption of technologies it does acquire, more systems engineers and product managers to help bring acquisition projects to fruition, and more technologists and designers who can smartly build, buy, and integrate cutting-edge technology.
Defense tech leaders often speak about “unleashing” the power of the market by cutting regulations. But the defense sector is an inherently distorted and flawed marketplace. It is defined by a single dominant buyer, significant barriers to access, and large capital requirements; no amount of deregulation or deference to Silicon Valley can correct for that. Rather than cede yet more ground to the private sector or design policy around an unachievable vision of self-administering contracts available to flashy but untested firms, Washington must grab the regulatory and institutional reins to create and manage competition within the limits of a monopsony market. It must negotiate and oversee contracts that extract the best performance from the private sector, encouraging the competition that Silicon Valley rightly claims is currently lacking. And it must decide when public in-house research, development, or production is the better investment for the future.
Further weakening government institutions on the basis of misconstrued history will have devastating long-term costs. The first and second questions in any acquisition decision must always be: Does this choice strengthen the force today and in 50 years? and, Does it respect the taxpayer? Defense tech leaders and the government will not always agree on the answers. But the government’s role is to advance the public interest, not to please its contractors. A capable and competent state is not the enemy of innovation and resilience but a precondition for them.
Comments
Post a Comment