There is a tendency in Washington to measure technological competition by asking who has the most advanced system at any given moment. That is understandable. It is also incomplete.The competition with China over artificial intelligence will not ultimately be decided by a benchmark or a product launch. It will be decided by something much larger: whose technology the world uses, whose platforms entrepreneurs build upon, and whose standards become embedded in the global economy.America remains in an enviable position. We have extraordinary computing capacity, the deepest technology markets in the world, exceptional universities, experienced entrepreneurs, and companies operating at the frontier of AI.
But advantages are only useful if we know how to exploit them.China is pursuing an aggressive strategy of making its AI models available to developers around the world. Chinese firms including Alibaba and DeepSeek understand that there is strategic value in adoption. If a developer in Brazil, a manufacturer in India, or a startup in Europe builds its business around Chinese technology today, China gains influence over the technological ecosystem of tomorrow.We should not underestimate that strategy simply because giving away technology appears, at first glance, to be bad business.It may be very good geopolitics.Open-source AI gives developers the ability to take a model, adapt it to a particular problem, operate it on their own infrastructure, and build new products without remaining permanently dependent on the company that originally created it. That makes these models attractive to startups and researchers, but also to manufacturers, hospitals, universities, governments, and other institutions that may need greater control over their technology and their data.The United States should want those institutions building on American models.And America’s innovators are the first to recognize this. In an unprecedented fashion, the country’s largest tech companies, including Microsoft, Palantir, and Nvidia, have recently launched a coalition to protect and promote open-source AI. The alliance is meant to “broaden the community of defenders by sharing models, tooling and research in the open.” I cannot think of another instance where Silicon Valley displayed such unity on a single policy issue.This is not an argument against proprietary AI. American companies invest enormous amounts of capital developing advanced models, and they are entitled to protect intellectual property and earn returns on those investments. Closed models will remain an important part of the AI economy.The mistake would be allowing the commercial preferences of individual companies to become the strategic policy of the U.S.A company has a responsibility to its investors. The U.S. has a responsibility to preserve American technological leadership.Those objectives frequently align, but not always.I learned during my time at the Department of Homeland Security that resilience rarely comes from concentrating capability in one place. Whether we are discussing critical infrastructure, supply chains, communications systems, or emerging technology, dependence creates vulnerability.That principle applies to AI.If access to America’s most capable AI systems becomes concentrated among a small number of corporations, we risk creating technological choke points in an industry that will increasingly touch virtually every sector of the economy. Small businesses become dependent on pricing decisions they do not control. Researchers are limited in what they can inspect. Companies building specialized systems remain tied to another firm’s infrastructure and rules.Open models provide an important counterweight.They allow an American startup to build without seeking permission from a gatekeeper. They allow manufacturers to tailor systems to their operations. They allow researchers to examine models rather than simply accepting their outputs. And they can allow sensitive organizations to deploy AI within controlled environments rather than sending valuable data outside their walls.There are security concerns associated with making powerful technology widely available, and dismissing them would be irresponsible. AI will create new capabilities for adversaries as well as allies. We should establish appropriate safeguards around genuinely dangerous capabilities and continue investing heavily in AI security.But there is a difference between addressing a specific security risk and embracing technological closure as a governing philosophy.America did not become the world’s leading technological power by being afraid of competition.Our advantage has always been the opposite. We create. We compete. We attract talent. We give entrepreneurs room to experiment. When somebody develops a better product, another American tries to build something better still.That culture matters enormously in AI because the technology is moving too quickly for anyone to know where the most important breakthrough will originate. It may come from one of today’s largest technology companies. It may come from a university laboratory. It may come from five engineers working at a startup nobody has heard of yet.Our job should be to make sure they are building in an American ecosystem.The strategic danger is not difficult to see. If American companies offer exceptional but expensive proprietary systems while Chinese companies offer capable models that are inexpensive, adaptable, and readily available, much of the world will choose the latter.Not because foreign developers prefer China.Because they prefer what is accessible.And once companies build products around a particular model, train employees to use it, create software compatible with it, and invest capital around that ecosystem, those relationships become difficult to unwind.That is how technological influence compounds.America therefore needs an AI strategy concerned not merely with developing the world’s best models, but with ensuring the world builds on American technology.That means maintaining our lead in computing infrastructure. It means encouraging competition rather than protecting incumbents. It means resisting regulations that unintentionally make it impossible for smaller American companies to compete. And it means recognizing open-source AI as a strategic asset rather than treating it as an inconvenience to established business models.We should approach the AI race with confidence. America possesses advantages China has spent decades trying to replicate. But history is full of nations that possessed advantages and squandered them because they mistook their current position for a permanent one.China wants its technology to become indispensable to the world.We should want the same for ours.BLAME POPULAR CULTURE FOR THE FEAR OF AIThe objective is not simply to build the best AI in America. It is to ensure that when the next generation of entrepreneurs, companies, researchers, and governments decide what technology to build upon, the answer is American.That is the competition that matters. And it is one America cannot afford to lose.Joe Guy served as deputy chief of staff to the secretary of homeland security under President Donald Trump.
