The Chinese artificial intelligence (AI) lab DeepSeek grabbed headlines and tanked the stock market with its announcement of a new AI model nearly equivalent to the United States’ most recent reasoning models but at a fraction of the cost. If the past is prologue, the DeepSeek development will be seized upon by some as rationale for eliminating domestic oversight and allowing Big Tech to become more powerful.
Such a response is exactly the opposite of what America needs. If anything, DeepSeek proves the importance of protecting American innovation by promoting American competition.
A typical Silicon Valley argument has been that allowing big companies to gobble up smaller rivals allows the incredible resources of those big companies to drive the AI race forward and protect American interests. Such a thesis conveniently overlooks that the breakthroughs of DeepSeek, OpenAI, and Anthropic were breakthroughs from disruptive startups, not national champions.
“We often hear that pursuing antitrust cases against or regulating [dominant tech] firms will weaken American innovation and cede the global stage to China,” former Federal Trade Commission (FTC) chair Lina Khan has said. But “history and experience show that lumbering monopolies…cannot deliver the breakthrough technological advancements that hungry startups tend to create…To stay ahead globally, we don’t need to protect our monopolies from innovation – we need to protect innovation from our monopolies. We need to choose competition over national champions.”
Not only are big companies lumbering but cutting-edge innovations often conflict with corporate interest. According to The Wall Street Journal, Google engineers had built a generative AI chatbot over two years before OpenAI unveiled ChatGPT. Despite entreaties from its engineers, Google management sat on the breakthrough. One theory is that the ability to ask an AI chatbot a question and receive an answer threatened Google search, the company’s cash cow. “Gmail creator warns Google is ‘only a year or two away from total disruption’ because of AI like ChatGPT,” one headline proclaimed.
An incumbent like Google—especially a dominant incumbent—must continually measure the impact of new technology it may be developing on its existing business. The company’s first fiduciary responsibility is to shareholders, not to the expansion of knowledge.
An innovative startup such as OpenAI, however, has no such qualms. “If it’s going to happen anyway, it seems like it would be good for someone other than Google to do it first,” OpenAI’s CEO Sam Altman wrote in an email to co-founder Elon Musk.
A strategy of “national champions” is at the heart of China’s AI policies. The Chinese government anointed big companies such as Baidu, Tencent, and Alibaba.Then, the Chinese government subsidized them with cash and helpful policies. Yet DeepSeek was built, not by these favored companies but by a hedge fund that originally started using AI for trading decisions.
The commoditization of AI
DeepSeek’s breakthrough took advantage of the commoditization of AI. There are over one million open-source models freely available on the Hugging Face open-source repository. These open-source models, built on breakthroughs in the original foundation models, are free to be modified and developed as the user sees fit.
DeepSeek studied those open-source models, trained their own model, and optimized it to use less computing power. Then, they open-sourced their breakthrough to make it available to everyone.
The evolution of AI from amazing proprietary capabilities to an openly available commodity is a watershed that will enable the proliferation of innovation, not just in the foundation models, but in the widespread application of the technology. The lesson of history is that it is not the primary technology that is transformative, but its secondary applications. The open availability of a low-cost, low-compute model opens the door to the Jevons paradox, an economic principle that increased efficiency leads to greater overall consumption rather than a reduction. As Microsoft CEO Satya Nadella posted on X after the DeepSeek announcement, “Jevons paradox strikes again!”
AI’s second lap
The AI race has now begun its second lap. Model development will continue to be important, but the future lies in what easily available AI will enable. As Jack Clark, co-founder of Anthropic, explained, “DeepSeek means AI proliferation is guaranteed.”
The dispersal of AI applications in the United States is driven by for-profit enterprises seeking to gain a competitive advantage. Its opportunities are virtually boundless in areas as diverse as healthcare, consumer tech, finance, or farming. This second leg of the AI race, however, requires the maintenance of an open marketplace environment that avoids innovations being gobbled up by the kind of market dominating power that characterized the last quarter century.
The Big Tech companies, often looked to as America’s national champions, have become big through anticompetitive activities. The Department of Justice and multiple state attorneys general sued Google for violating antitrust laws to dominate the search market (and won.) They also sued Google’s online advertising market and expects a decision soon. The FTC has sued Meta alleging an unlawful effort to maintain their social media monopoly by acquiring Instagram and WhatsApp, a case that is expected to go to trial in April.
If we are concerned about the AI race with China, we need to focus less on lobbying to let the big guys get bigger, and more on making sure there are competitive opportunities to spur innovation.
Promoting innovation
As today’s AI developers mature and as AI disperses into applications, the historical lesson remains critical: unchecked consolidation of power stifles the innovation necessary for economic growth, national security, and consumer protection. Ensuring a competitive market drives innovation. This includes not only antitrust enforcement, but also sectoral regulation built on promoting competition while providing consumer protection guardrails.
Innovative competition also requires support for the innovators. The venture capitalist model predicated on the sale of the startup to a dominant company is broken. Investing with the goal of ultimately consolidating the new competition into existing powerhouses may maximize VC returns, but does not maximize returns to the public interest.
Building the competition necessary for a vibrant AI market requires alternative support vehicles for innovators. Such support could include initiatives by the Small Business Administration and tax policies. At the same time, easing the path for initial public offerings could provide an alternative exit strategy for those who do invest.
Innovation proliferation also proliferates the risks of existential harm from unsupervised AI. That a hedge fund can revolutionize the AI world seemingly overnight highlights the importance of assuring the headlong rush forward includes behavioral expectations and guardrails. President Trump’s ill-advised repeal of President Biden’s executive order removed some of the safeguards against runaway AI. Positive AI developments require balancing open source technology with safety standards and the enforceable expectation they will be followed. This includes red teams to actively seek problems in new models and report their findings. Such standards and behaviors require audits at least to the level we require for the financial system.
Ultimately, the scare headlines that a new Chinese AI model threatens America’s AI dominance is just that—a scare headline. If appropriately applied in a competitive marketplace, the DeepSeek development is a roadmap for continued American leadership. Built on U.S. technology, it commoditizes AI and accelerates the race to to disperse AI throughout the economy of the world.
Such threats and promises require oversight. A competitive market that will incentivize innovation, must be accompanied by common sense guardrails to protect against the technology’s runaway potential. The arrival of DeepSeek shows that competition works; it represents an opportunity for the United States to continue its AI leadership.
