To understand what went wrong with antitrust during the Biden administration, look no further than former Federal Trade Commission (FTC) chair Lina Khan’s take on DeepSeek’s launch of R1, an artificial-intelligence (AI) platform. Rather than see the moment for what it is — a competitor arising by bypassing Biden’s supposedly insurmountable barriers to competition — she sees this as evidence that America’s tech companies are monopolies in need of regulation. Talk about willful blindness.

The Chinese company’s R1 sent shockwaves through the tech world. The quality of R1’s outputs has been widely praised, its costs are reportedly a fraction of those of leading AI companies, and it was developed in record time. Its rapid rise to the top of app-store charts, both in the U.S. and overseas, stunned the industry.

This is economist Joseph Schumpeter’s “creative destruction” in action but not at its best. The platform represents a leap forward in certain respects, but it also reportedly exploited loopholes in American AI platforms. R1 also exposes holes in the Biden team’s antitrust theories.

For years, Khan and her allies argued that only they could take down Big Tech and create competition in digital markets. They believe that Alphabet, Amazon, Apple, and Meta maintain dominance in the marketplace because new entrants cannot possibly replicate the vast resources — data, algorithms, and computing power — needed.

DeepSeek’s achievement reveals the folly of these regulators’ beliefs. But rather than recognize the competition that everyone from Wall Street to Silicon Valley sees, Khan allows her thinking to be shaped by confirmation bias and motivated reasoning, falsely embracing the incoherent conviction that even though, as she says, “our big tech firms are at risk of being surpassed in A.I. innovation by foreign competitors,” there “isn’t enough competition.”

These beliefs took both the Biden administration and the European Union (EU) down an antitrust rabbit hole. The EU adopted regulations, including the Digital Markets Act, to control how Big Tech companies use AI. Meanwhile, President Biden issued a sweeping executive order aimed at prescribing how AI should be developed and deployed. His administration wanted to break up Meta and Alphabet, and even proposed restricting Alphabet’s future AI efforts. These cases are still in play, creating a need for President Trump to put antitrust back on its proper footing.

DeepSeek’s R1 is a case study in how market incentives — not government — spark ingenuity to overcome barriers. When critical resources become constraints, innovators create alternatives. The conventional wisdom had been that AI innovations required immense scale: billions of dollars in research, vast data troves, hyperscale data centers, and access to cutting-edge chip technology. DeepSeek’s engineers developed new algorithms and exploited loopholes in existing platforms to sidestep technology restrictions and reduce training costs.

These tactics highlight a fundamental reality: Barriers that seem insurmountable are always temporary. The antitrust cases against Big Tech rest on the premise that it is prohibitively expensive for rivals to acquire the resources and scale needed to compete in search, e-commerce, social media, and digital advertising. DeepSeek’s rapid ascent proves otherwise.

This has profound implications for the tech industry. American AI companies must rethink their business models. While fundamental AI advancements still require significant computing power and research, DeepSeek’s innovations and the rise of model distillation — essentially unauthorized use of another platform’s work — allow competitors to piggyback on others’ innovations.

Outside of the business world, antitrust regulators must reconsider their approach. Breaking up companies chokes their AI engines. And giving their resources to rivals, as the Biden team proposed, would encourage more exploitation while stifling fundamental breakthroughs. Khan is correct in her recent opinion piece that Alphabet failed to exploit its Transformer AI architecture, but she is wrong that large companies don’t innovate. Academic research consistently shows that new creations come from large and small companies alike. Any company can have blind spots, as can regulators.

DeepSeek’s success is a reminder that markets are dynamic, and innovation comes unexpectedly and from unexpected sources.

The launch of DeepSeek’s R1 should serve as a wake-up call for antitrust authorities. Economic realities are evolving rapidly and clinging to outdated assumptions risks stifling the very innovation regulators claim to want. Creative destruction is alive and well — and it’s always coming for the status quo.