With major developments taking shape at a rapid pace in artificial intelligence (AI), including DeepSeek’s $6 million breakthrough, a new reality is dawning for tech markets. Experts and investors take away a variety of lessons, from the straightforward conclusion that more is now possible with less in AI—to demonstrably inaccurate takes, such as that there is insufficient competition in AI. The latter view dominated the Biden Administration’s antitrust approach to AI markets, and early indications suggest “Big Tech” is still in the federal government’s crosshairs under President Trump. However, AI optimism may yet overtake the more interventionist inclinations inherent in the pervading populist agenda, suggesting small businesses in these markets may benefit from an era of growth and the flexibility to innovate on robust platforms. Perhaps most heartening are the views articulated in Vice President JD Vance’s speech this week in Paris, where he argued that “to restrict [AI’s] development now will not only unfairly benefit incumbents in the space, but it would also mean paralyzing one of the most promising technologies we have seen in generations. . . . The development of cutting-edge AI in the U.S. is no accident. By preserving an open regulatory environment, we’ve encouraged American innovators to experiment and make unparalleled R&D investments.”
Here, the VP talked mainly about AI-specific regulations and liability regimes that the European Union (EU) is considering. However, his comments apply similarly to interventions rooted in competition or antitrust law and policy. In the United States, federal officials have recently struggled to understand the meaning of advancements in AI and what to do about the potential risks they present. In the process, they also underappreciate the benefits of investments in AI while overestimating the risks they pose to competition.
More than its demonstration of technical progress, DeepSeek vividly illustrates the triumph of innovation over constraints—be they access to inputs, availability of capital, or run-of-the-mill regulatory hurdles—when markets are truly competitive. But leave it to the government (and, to be fair, investors) to never have seen something like this on the horizon. The Federal Trade Commission’s (FTC’s) report on Partnerships Between Cloud Service Providers (CSPs) and AI Developers is one example of how federal officials have tended to underestimate relatively under-resourced competitors and overestimate the risks to competition of investments by large CSP incumbents. Although the FTC’s “areas to watch” in Section 5 of the report are not binding regulations or enforcement actions, they appear to try and undermine investment by CSPs by focusing on potential competition constraints that are increasingly unlikely to create antitrust problems as events unfold in AI markets.
The report warns that CSP partnerships with AI developers could “affect access to certain inputs, such as computing resources and engineering talent.” Specifically, Microsoft and OpenAI, for example, agreed that Microsoft would have exclusive rights to build computing capacity for OpenAI (subject to caveats) in exchange for billions of investment dollars. Seeing a smaller player in OpenAI making a real tradeoff by conceding exclusivity to Microsoft, the FTC perceived a need for the government to ride to the rescue, even if only via threatening subpoenas issued under its 6(b) authority. Ultimately, this seemingly permanent exclusive arrangement turned into a “right of first refusal” for Microsoft, allowing OpenAI to look elsewhere without technically being in breach. Simultaneously, DeepSeek upended input cost assumptions by providing roughly equivalent services to OpenAI’s GPT-4 at for 32.8 times cheaper than what OpenAI invested.
These events have illustrated that we have no idea what effect the drastic leap in computing efficiency will truly have, except that competitors will jockey as swiftly as possible to capitalize if allowed to do so. Summarizing all of these unknowns by focusing only on the ways competitors could restrain future competition needlessly steers policymakers toward intervening in these markets to remedy potential, possible (but rather unlikely) future market failures.
Despite a lack of access to the latest chips and infrastructure or the capital necessary to even pay for them if they were available, DeepSeek was not well positioned to succeed and did so despite its predicament. The development challenges AI antitrust hawks’ assumptions. In awe of the size and power of large CSPs, the FTC appears to have forgotten that their efforts to secure exclusivity are readily assailable as efficiency advancements and other market realities shift the foundations of our collective assumptions. Incoming FTC Chairman Andrew Ferguson concurred in part with the decision to issue the report but dissented with respect to Section 5. The “limited, brief nature of the study should foreclose the drawing of broad conclusions about the AI industry and its future . . ..” Ultimately, a view like this admits that the government is not omniscient, presaging a return to former FTC Acting Chairman Maureen Ohlhausen’s “regulatory humility” ethos. If it extends broadly to the federal government’s antitrust philosophy in tech markets, especially on AI, it is good news for small businesses in the app economy.