Join War on the Rocks and gain access to content trusted by policymakers, military leaders, and strategic thinkers worldwide.
Editor’s note: This is the first article in a limited series celebrating American defense technologies born from wartime and their effects on broader national security, politics, and society. This series will run for several weeks to commemorate America’s 250th anniversary, and winners will be selected by a reader vote undertaken through our newsletter later this summer. Prior installments can be found at the Arsenal of Innovation page.
The history of the semiconductor is an origin story for modern computing but also reveals a recurring pattern in American innovation: government helps underwrite technological breakthroughs, and commercial markets transform them into general-purpose technologies. And yet, paradoxically, the American innovation model can become a victim of its own success. The institutions that help launch transformative technologies often lose influence over them as they become commercially indispensable. The semiconductor story therefore provides a useful lens for understanding the opportunities and tensions now emerging around artificial intelligence.
That story began in the summer of 1958. Jack Kilby, an electrical engineer, was just starting a new job at Texas Instruments in Dallas. His task, funded by the U.S. Army, was to miniaturize electronics by stacking thin ceramic wafers into modules. Kilby saw the approach as a dead end because it required a complex circuit of soldered joints. Each one represented a potential failure, like a string of old Christmas lights in which one malfunction meant the whole strand would not turn on.
In an autobiographical aside within an electrical engineering article on the origins of the integrated circuit, Kilby recalls that “I had no vacation time coming and was left alone to ponder the results of the IF [intermediate frequency] amplifier.” During this period, he developed a new approach based on integrating semiconductor elements into a single circuit. He sketched a small germanium bar with hand-wired gold leads, built the device, and, when his colleagues returned, powered it on and observed a sine wave on the oscilloscope, demonstrating the feasibility of the integrated circuit. Kilby had solved the functional problem, but the cost was prohibitive for commercial applications.
For the Air Force, cost was not the primary constraint. Air Force officials warned that “interconnection failures seriously threatened the accomplishment of even routine operational tasks.” During the Cold War, increasingly sophisticated weapons and communications incorporated thousands of electronic components, and reliability became a critical concern. Every transistor, resistor, capacitor, solder joint, and wire introduced a point of failure.
Under Air Force sponsorship, Texas Instruments demonstrated the potential by constructing two functionally equivalent computers: one with 9,000 discrete electronic components and another using 587 integrated circuits. Each integrated circuit contained many individual electronic elements, but reducing the number of separately manufactured and interconnected packages by more than an order of magnitude greatly simplified assembly, improved reliability, and reduced the size and weight of electronics. Convinced that the new approach could overcome “the tyranny of numbers,” the Air Force funded Texas Instruments to produce integrated circuits at scale.
Those early defense purchases not only solved a technical problem but also created a stable source of financial support in the market, enabling firms to refine manufacturing, increase production volumes, and further drive down the costs, which laid the foundation for the commercial diffusion of integrated circuits. Within a decade, Jack Kilby and his team had developed a hand-held calculator built on integrated circuits. Four years later, Intel, founded by engineers who had left Fairchild Semiconductor, introduced the Intel 4004, the first commercially available single-chip microprocessor. Originally developed for a Japanese calculator manufacturer, the 4004 condensed the central processing unit of a computer onto a single chip, laying the technological foundation for the personal computer revolution.
What followed was one of the most remarkable periods of technological progress in modern history. In 1965, Gordon Moore, who co-founded Intel, observed that the number of components that could be placed on an integrated circuit was doubling at a regular rate, while costs per component continued to fall. The prediction, later known as Moore’s Law, became the organizing principle for the semiconductor industry. Firms invested continuously in new fabrication techniques, manufacturing processes, and chip architectures to sustain the pace of improvement. The result was a relentless cycle of innovation in which each generation of chips became smaller, faster, cheaper, and more capable than the last. What began as a niche military technology evolved into a general-purpose technology that transformed virtually every sector of the economy, from communications and transportation to finance, healthcare, and national defense.
The experience of the chip from risky investment to a ubiquitous, indispensable feature of the economy is a case study in American innovation. It involved the resources of a public institution, in this case the Air Force, and the ideas and initiative contained within a private corporation. The engineer was first trying to solve a physics problem that had stood in the way of missile guidance, but the innovation generalized far beyond that, transforming commercial electronics.
The partnership between public and private institutions is one of the most defining features of American political economy. But as the case of the chip and now AI show, it also reflects tensions that today are stress-testing the same model.
The process that produced ubiquity also produced dependence. As semiconductor manufacturing became specialized and globally distributed, efficiency and innovation came at the cost of domestic industrial resilience. The COVID-19 pandemic exposed the fragility of globally distributed semiconductor supply chains as chip shortages rippled through both commercial and defense sectors. China’s emergence as a strategic competitor further transformed semiconductors from a commercial input into a contested asset of economic and national security policy. At the same time, AI has concentrated demand on the most advanced chips, many of which depend on fabrication capacity concentrated in Taiwan.
Across both the Biden and Trump administrations, U.S. semiconductor policy has centered on leading-edge chips because the overwhelming majority of their manufacturing capacity is concentrated in Taiwan, beyond direct US control, and because they have become a critical bottleneck for advanced AI.
Some of the bigger vulnerabilities, however, are more mundane, located in the more mature, lagging-edge chips. During the pandemic, shortages of microcontrollers in cars, industrial machinery, and even defense systems forced production lines of these chips to shut down. Further, many defense platforms rely on older-generation chips that are not manufactured at scale and are sourced from secondary markets where counterfeit or compromised components can enter the supply chain. The vulnerability is therefore both availability and whether these chips can be trusted.
The extent of these concerns is reflected in recent American industrial policy. Beyond subsidies and tax incentives, the federal government took an approximately 10 percent equity stake in Intel in 2025. This is part of a broader effort to preserve domestic semiconductor manufacturing capacity and secure access to advanced chips for national security applications. Although the semiconductor industry has long benefited from defense procurement and public investment, the scale of direct government intervention marks a departure from the market-oriented approach that characterized much of the globalization era. The move highlighted a broader shift in the relationship between states and markets.
During the Cold War, the government’s role as a major customer gave it significant influence over the direction of technological development. The commercial semiconductor market now dwarfs its original defense customer. The information technology revolution transformed semiconductors from a niche military experiment into the foundation of a global commercial industry optimized for efficiency, cost, and shareholder returns. While those firms pursued economically rational decisions, the United States became dependent on overseas manufacturing capacity for technologies that it now regarded as strategically indispensable.
The result is a paradox. The partnership that helped create one of the most successful technologies in history also produced strategic challenges that neither government nor industry can address alone.
The semiconductor story was about the government losing influence over a technology it helped create. AI pushes that trend even further. Whereas government once shaped technological development, commercial markets drive the frontier.
The comparison is not exact. As with semiconductors, the federal government played a foundational role in AI by funding decades of basic research. Unlike the integrated circuit, however, frontier AI capabilities have been developed primarily within private firms rather than through defense procurement. AI introduces a twist on the old dilemma. In the semiconductor era, the government helped create a defense technology that eventually became essential to both civilian prosperity and military power. In the AI era, the relationship has been inverted. Frontier capabilities have emerged primarily from civilian firms, while the government finds itself in the position of customer rather than sponsor.
The private-sector-led AI ecosystem has advantages. Competition among firms has accelerated innovation by attracting enormous pools of talent and capital, producing capabilities that government laboratories would have struggled to replicate on their own. Yet the shift alters the balance of influence between the state and industry.
Those capabilities have become national security assets integral to intelligence, cyber operations, and military planning and operations. Decisions that in a different domain might be a narrow matter of corporate governance regarding access, safeguards, and capabilities can quickly elevate to matters of national security. As the systems have become more capable, the line between private innovation and public interest has become vanishingly small.
Tensions have already begun to emerge. For example, a company like Anthropic is attempting to cultivate a reputation for ethical and responsible AI. Chief Executive Officer Dario Amodei has consistently warned about AI’s security implications and sought to impose limits on access to its models and on the purposes for which they are employed. Almost as if heeding those warnings, the Trump Administration began placing export controls on advanced models on national security grounds, citing risks of proliferation to adversaries, potential misuse in cyber operations or weapons development, and the need to maintain a U.S. technological edge. It later retreated on some of those restrictions, but the episode pointed to a historical irony: The government, which has long encouraged markets to innovate, now felt the need to constrain access to these technologies. The trajectory suggests that frontier AI may have become too strategically important to release in the same way advanced semiconductor manufacturing became too strategically important to offshore.
The integrated circuit is a story of technological ingenuity. But it is just as much a story about institutions. An Air Force procurement decision helped transform an expensive and unproven military technology into a foundation of the modern economy. Yet the same process that produced extraordinary innovation also created new dependencies, new vulnerabilities, and new challenges about the relationship between public authority and private power.
That tension is not unique to semiconductors. It is reappearing in artificial intelligence and will likely reappear in future strategic technologies. The lesson of the chip is therefore not just that defense investments can generate transformative innovations across society. The institutions that nurture innovation should also be capable of managing its consequences.
The integrated circuit ultimately reveals three enduring challenges of American innovation. The first is the importance of risk underwriting. The Air Force did not know that the integrated circuit would become the foundation of the modern economy. What it did know was that the technology offered potential military advantages worth pursuing despite its high cost and uncertain future. Commercial markets were not yet prepared to absorb those risks, but the Air Force was. In doing so, it helped create the conditions under which private firms could experiment, scale production, and eventually drive costs down. The lesson is that governments with sufficient resources and institutional capacity can absorb risks that markets are unwilling to bear, creating opportunities for innovation that would otherwise never reach scale.
The second is continuity. Semiconductors evolved over decades, through changes in administrations, military priorities, and market conditions. AI will likely do the same. Elections should have consequences, but technologies that shape national power cannot be governed exclusively through four-year political cycles. The challenge is building institutions capable of distinguishing enduring national interests from the preferences of any particular administration, regulator, or business leader.
The third is partnership. The history of the chip demonstrates that American innovation has never been solely public or private. It emerged from the interaction of government demand, private initiative, scientific talent, and entrepreneurial ambition. That partnership can create tensions, dependencies, and occasional conflict. Yet it is also the source of much of America’s technological dynamism. It’s critical to manage those tensions before they corrode the alignment between national interest and private incentives.
China’s system earns plaudits for being coherent and focused. While the American system of innovation is messier, the story of the integrated circuit suggests that some of that messiness is a feature rather than a flaw. The semiconductor revolution emerged from the interaction of military necessity, entrepreneurial ambition, and public investment. The process was not always efficient either along the way or in its recent experience, but was generally adaptive.
Neither the Air Force officers who approved the initial semiconductor procurement contracts nor the engineers who designed the earliest chips could have foreseen smartphones or AI. They built a system capable of underwriting risks and translating military necessity into broader technological progress. The American advantage is not about choosing between markets and the state. It is its dynamic ability to rebalance the relationship between the two as technology changes.
Sarah Kreps is the director of the Tech Policy Institute and a professor at Cornell University. She is the author of eight books, including the forthcoming Harnessing Disruption: Building the Tech Future Without Breaking Society (Oxford University Press). She previously served as an officer in the United States Air Force.
Image: Simon Claessen via Wikimedia Commons.
