When President Trump proclaimed a new “golden age” for America during his inauguration, he did so with a platoon of tech titans literally and figuratively behind him. He appears ready to unleash these alt-neu techno-optimists by rekindling the latent energy of the free market and, in particular, supercharging America’s AI development. In February, Vice President Vance told European grandees gathered at  the Paris AI Summit that “the AI future is not going to be won by hand-wringing about safety. It will be won by building.” 

Accelerating innovation is more than a domestic imperative — it is perhaps the most crucial foreign policy choice of our time. The United States faces a great-power competition with China that is, at its root, a technology competition, with fierce battles for supremacy  across domains ranging from frontier AI models to lightweight batteries..  In that respect, the U.S.-China contest appears to be a 21st-century version of past rivalries in which technology has proven crucial to national power, from the struggle between Great Britain and Germany in the leadup to World War I to the Cold War. 

But many have yet to grasp the new velocity of technological change, and how it’s transforming great-power competition. In the past, the geopolitical consequences of breakthrough innovations often took decades to permeate and reshape world order. Aviation, for example, took decades to confer a decisive advantage in war, giving early laggards ample time to catch up. Foundational technologies of the industrial revolution, such as railroads and electric power, proliferated only as fast as operators could build out capital-intensive networks of steel rails and copper wire.. Such innovations developed and compounded slowly, layer by layer, constrained by physical and financial limits. Yet digital technologies, such as AI, compound exponentially. New large language models are emerging at a torrid pace--between December 2024 and April 2025, for example, xAI, Anthropic, OpenAI, Google, and DeepSeek all launched new models--which in turn drive rapid advances in biotechnology, robotics, and manufacturing. In this new world, Tech develops faster because each advance unlocks the next, cycles compress as knowledge grows, and falling behind risks locking in deficits. Missing even one innovation cycle can thus shift the global balance of power. 

Keeping technological pace will prove critical to U.S. national security. The nation that wins the AI race will obtain a decisive edge, securing a future of rapid and growing economic, military, and cultural dominance -- a kind of geopolitical flywheel. Should China prevail, it will likely reshape world order through domination, wielding its technological advantage to entrench Community Party rule by crushing dissent, hoarding resources, accruing wealth, and enforcing global subservience to its whims. A China-shaped world, in other words, would devastate Americans’ way of life, draining their prosperity and choking their freedoms.

Embracing the creative destruction of capitalism is thus not merely good economic policy. For all the upheaval new technologies can bring — and the understandable temptation to manage or slow innovation to reduce that upheaval — letting innovation rip is our only plausible strategy to prevail against China. It is the prerequisite to winning, and indeed, to running the race at all.  

✺

Even for a great power, failing to adopt a technology supercycle — in this case, the software revolution — can send it tumbling down the global-power leaderboard. In living memory, Europe’s leading economies were as wealthy as the United States. In 1980, for example, France’s GDP per capita exceeded that of the United States. Yet as U.S. technology firms surged, Europe prioritized regulation over innovation, boasting of its prowess as a “regulatory superpower.” In the past decade, the European Union has spawned regulations meant to safeguard consumer privacy (the General Data Protection Regulation ) , prevent the formation of tech monopolies (the Digital Markets Act), and reduce risks from runaway AI (the AI Act). 

These rules dampened growth and innovation. By mandating cumbersome data handling, strict rules on interoperability, cumbersome audit and safety requirements, and high taxes, Europe’s laws drove up compliance costs, discouraged risk-taking, slowed iteration cycles, and scared off investment and talent. The result: Thirty percent of unicorn startups founded in Europe since the 2008 financial crisis relocated abroad, mostly to the United States. Meanwhile, seventy percent of the top technology companies in the world are American. The effects of Europe’s lost decades on the relative power of the two continental blocs are striking. Today, U.S. GDP per capita is nearly double that of France.  Even Mississippi, the poorest U.S. state, has a per-capita GDP greater than four of five of Europe’s leading economies, with only Germany’s narrowly higher.

While Europe missed the latest technology cycle, China seized it. Beginning in the 1980s, China adopted a model of centralized capitalist planning, pouring government money into important hand-picked industries––often with civilian and military applications. China also had no qualms about “borrowing” American innovations for its own companies; sometimes legally, through lopsided “joint ventures,” and sometimes illegally, through hacking and industrial espionage. Under this state-backed umbrella, Chinese entrepreneurs have built a formidable low-cost manufacturing base, ringing up a $1 trillion trade surplus through state subsidies and a weakened yuan. They also built a legion of innovative companies, including DeepSeek — a frontier AI model whose impressive reasoning model, with technology reverse-engineered from OpenAI, wiped $1 trillion from the market caps of American AI incumbents in one day. 

Beijing succeeded in part because its centralized approach allows it to move quickly, pouring resources into short, explosive tech cycles to seize control of them both upstream, through rare earths and raw materials, and downstream, through cheap, mass-manufactured goods. It also artificially subsidizes unprofitable businesses until they can capture markets with underpriced goods and advantages its own companies over foreign competitors. China has also successfully dodged American attempts to throttle this strategy, such as chip bans and trade curbs. Its fusion of speed, scale, and top-down control have combined to catch up with the West in some sectors once thought impregnable, such as AI, and out-innovate in others, such as EV battery technology. 

✺

To be sure, China’s top-down innovation strategy is not without costs.  Its comprehensive social control can dampen innovation and deter investment, as entrepreneurs and investors shun new ideas that could displease the Party. A 2020-21 crackdown on corporations, beginning with pulling Alibaba’s IPO, wiped more than $1 trillion worldwide from the market value of Chinese companies. And the CCP hardly has a flawless record of steering capital. Late last decade, for example, the CCP launched a series of multi-billion dollar semiconductor projects meant to rival TSMC and Samsung. All failed, with many billions reportedly stolen or squandered.  

The United States, by contrast, possesses the most fruitful innovation market in the world. Silicon Valley’s freewheeling dynamism, venture capital, and, above all, devotion to meritocracy is peerless (there is a reason why Elon Musk left South Africa to build rockets here, not Beijing). Low-friction talent sorting lets ambitious minds flow to hubs like San Francisco or Austin, unhindered by bureaucracy or borders. And comparatively modest  red tape and high upside encourage risk taking and competition, ensuring that the strongest and most innovative companies emerge.

The United States can press its advantages to exploit China’s weaknesses. The faster we push ahead, the more the CCP may need to loosen the reins on its entrepreneurs to keep up––which may compel the Party to ease its control over its own people. Our system is flexible enough to absorb the bumps of dramatic social, economic, and technological change; for the CCP, such bumps could undermine the regime’s legitimacy, posing an existential threat to its rule.

✺

Despite the advantages that China enjoys in aligning resources and incentives with strategic goals, the United States remains more nimble, adaptive, and disruptive. We can out-innovate China over time by drawing on that inherent dynamism–if we have the courage to embrace change. 

Some federal and state policymakers, however, have come to view that dynamism as a liability. Some (call them tech Sinophiles), simultaneously admire and fear China’s rise in key technologies.  They would have the United States embrace a more state-driven approach to innovation to ensure our competitiveness. Others (call them tech Europhiles) argue that the potential benefits of building AGI are not enough to justify the risks.  This group envies Europe’s safety-first approach to new technologies.

Fantasies of becoming “China for a day” have long beguiled a small but vocal faction of America’s elite.  Most recently, Chinese advances in 5G and AI convinced many in Washington that the United States needs its own version of federally subsidized innovation. The Biden administration, for instance, channeled hundreds of billions of dollars to the private sector through the the Inflation Reduction Act and the Infrastructure Investment and Jobs Act, aiming to rebuild U.S. industry from the top down. To be sure, federal funding for fundamental R&D, such as support for universities and national labs, undergirds much private sector innovation. Federally funded research fueled the post-World War II innovation boom: DARPA, for example, planted the earliest seeds of the internet, while NIH funding supported medical advancements that private investors, driven by short-term market incentives, might have otherwise ignored. But government’s useful role typically ends at the frontiers of research.  Moreover, the open, freewheeling, decentralized nature of the U.S. system means that we’re unlikely to beat China at its own central planning game. Our system of innovation thrives on speed, experimentation, and rapid iteration—qualities that government oversight inherently (and sometimes necessarily) stifles. It is one thing for SpaceX to launch hundreds of space flights, watching many of them fail and refining along the way; it is quite another for NASA, which fears the public backlash of a single failure, to do the same. China is a top-down autocracy that can force outcomes and smother failure and dissent. The United States excels when, to borrow a term, it lets a thousand flowers bloom, in fertile regulatory and R&D soil laid by the government.

While some argue for supercharged industrial policy, others argue that the risks of unshackled AI development outweigh the benefits, and look towards Europe for answers. This instinct spurred President Biden’s 2023 executive order on “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.” The order, ostensibly meant to strengthen America’s AI hand, focused largely on mitigating potential risks from AI: the order required developers of frontier AI models to file “safety” reports with the federal government and tasked federal agencies with proposing dozens of new AI regulations. 

Regulatory controls on AI like those in the Biden order enjoy vocal and influential support.  Skeptics behind these precautionary restrictions, such as the AI researcher Eliezer Yudkowsky and a plethora of AI safety think tanks, view AI, and, particularly, Artificial General Intelligence (AGI), as an existential threat to humanity. In 2023, for example, Yudkowsky contemplated airstrikes on “rogue” data centers being used to train  AI models. He and others fear that AGI will spiral out of control, destabilize society, or, in the worst case, eliminate the human race. 

But it is far from certain that machine intelligence will have the same aggressive, acquisitive drives as humans and other biological life forms.  Nor is it self-evident that a rogue AI could manipulate the ultimate complex system–human civilization–with the ease that AI doomsayers predict. 

Meanwhile,  slowing innovation in the name of safety carries its own risks. AI carries the potential to transform myriad industries, from biotechnology to manufacturing . Slower AI development means slower progress in those sectors–fewer diseases cured, fewer accidents avoided by autonomous driving–as well as slower GDP growth overall. 

After the 2024 election, the risk of overweening federal regulation has faded, for now. Three days into his term, President Trump revoked the Biden executive order, replacing it with his own, far shorter order calling for the United States to achieve “AI dominance.” With its light-touch approach, the Trump Administration is banking on the animal spirits of America’s tech investors and innovators to achieve that goal. 

Yet activists seeking to over-regulate AI have not given up: state governments are the next target. In particular, activists are pressing states to enact legislation putatively aimed at preventing “high-risk” AI systems from committing “algorithmic discrimination.” 

Researcher Adam Thierer dubs these laws “preemptive Precautionary Principle-based permitting regimes,” because they require innovators to  jump through various legal hoops before putting a model to work.  Those hoops can include “impact assessments,” “high-risk reports,” and “risk identification and management policies”--none which will usefully elucidate how the models actually work, but which will generate plenty of billables for the lawyers and consultants tasked with writing them.  The innovator’s gauntlet wouldn’t end there: once deployed, a model could then be faulted for producing “discriminatory” outcomes.  That can only mean disparate impact (Colorado’s law, for example, refers to “unlawful differential treatment or impact); machine-learning models can’t formulate discriminatory intent.  The problem is that no decisionmaking process, whether AI or human, will produce results that perfectly align with demographic statistics for every sensitive category.  (Unless, that is, demographic balancing is made the determinative factor.)  The result: if the regulation doesn’t kill AI breakthroughs, the “anti discrimination” litigation will.

In Texas, of all places, a proposed “Responsible AI Governance Act” would have “taken a risk-based approach to governance modeled after the EU AI Act.” The Act, like other such laws, would have imposed various regulatory burdens on AI developers and companies that use AI models for applications deemed “high-risk.” Fortunately, legislators heeded intense pushback from innovators and free-market advocates: a revised version of the bill drops references to “high-risk” AI and shears off the draft’s many bureaucratic and procedural hurdles to AI innovation.

Virginia had a similar near-miss.  In March, Governor Glenn Youngkin prudently vetoed a tellingly similar bill to regulate “high-risk” AI in order to prevent–there it is again–“algorithmic discrimination.”  Governor Youngkin explained that the bill would “harm the creation of new jobs, the attraction of new business investment, and the availability of innovative technology in the Commonwealth of Virginia. If economically significant states choke off AI development with overzealous regulation, even the best-intentioned federal plan will struggle to achieve “AI dominance” as China races ahead.

✺

There are risks to unfettered innovation. First, AI and its peer technologies will undoubtedly cause economic disruption, potentially at a breakneck pace, leaving little time for necessary adjustments. That disruption could cause social upheaval, fueling resentment against AI’s trailblazers. And while fears of rogue (or catastrophically inept) AGI may be overstated, they are not irrational.  

The problem is that there is no truly “safe” path.  Advocates of control present a choice between reckless, anarchic development on the one hand and safe, regulated AI research on the other. In a world without external competitors, perhaps so. But that world is not ours.  The seeming “safety” of going slow in fact presents another, unacceptable risk: that of finding ourselves under the thumb of an adversary whose power swells alongside its unbridled technological ambitions. Western handwringing will not stop China from pressing ahead, and if it seizes the commanding heights of 21st-century innovation, its lead will grow exponentially.  Indeed, even Europe seems to be emerging from its 20-year regulatory fugue, with the sacred cow of privacy regulation, GDPR, set to be trimmed in an effort to spur Europe’s tech sector.

In his History of the Peloponnesian War, Thucydides tells of a famous encounter between Athenian envoys and their counterparts from the tiny island of Melos. Demanding Melos’ surrender, the cynical Athenians remind the Melians of a hard truth of international affairs: “The strong do what they can, while the weak suffer what they must.” In today’s global race for technological dominance, “slow” makes “weak.” Nations that grow and innovate will lead in the future; nations that stagnate will be pulled along in their wake. 

The good news is that America is better equipped than any other major power to prevail in a competition driven by technology and innovation. Our chaotic, risk-loving culture—forged by Silicon Valley creativity and private sector ambition—can still outrun China’s brittle authoritarianism and Europe’s cautious overregulation. The great insight the current VC-to-DC movement can bring to Washington is to recognize that letting go of the reins has often been America’s winning competitive strategy. The Silicon Valley story reminds us that playing to America’s strengths often entails simply freeing things to run in unpredictable directions. In the coming technology competition, letting it rip is playing to win.

Adam I. Klein leads the Robert Strauss Center for International Security and Law at the University of Texas at Austin. He previously chaired the federal government’s Privacy and Civil Liberties Oversight Board, which oversees U.S. intelligence agencies.

Jordan Chandler Hirsch is a Senior Counselor at Palantir Technologies and a Senior Fellow in the Technology, Security, and Global Affairs Program at the Robert Strauss Center for International Security and Law at the University of Texas at Austin.