This week’s monopoly round-up has lots of fascinating news, both good and bad. Some serious quality trolling of Bank of America and JP Morgan by bank regulator Rohit Chopra, plus new FTC Chair Andrew Ferguson has a rough first week and The Economist gets actively corrupted by big tech money.

But first I want to highlight the emergence of Chinese artificial intelligence firm DeepSeek, which built a bunch of artificial intelligence models that are more cost effective than US models and recently released them on an open source basis. I don’t know the technology, but from what I understand, DeepSeek, instead of building out more raw computing power to train its model, focused on efficiency, and did so far better on that score than any U.S. company.

The U.S. has bet massively on this new technology, and China is prohibited from getting the best chips by American export controls. So the feat is more impressive than it appears. And it has shocked everyone in the space.

There are also broader consequences. Much of the American stock market valuation is based on the promise of artificial intelligence, where Microsoft/Amazon/Meta/Google, aka “hyperscalers,” are investing hundreds of billions of dollars to buy more data centers, and even rewriting the electric grid, including restarting a Three Mile Island nuclear reactor, to provide the juice to do so. The idea is that these huge investments in AI will provide a competitive advantage for American big tech to remake the world, and thus, for America to dominate.

DeepSeek’s accomplishment is thus terribly embarrassing, and also threatening for big tech and the American security apparatus. The presumption was that raw computing power is a competitive barrier in AI, these new models suggest it may not be as much as we thought. Meanwhile, short-sellers are salivating.

Of course, anything around AI and national security is foggy because it’s so embedded with the spying apparatus, with propaganda that is hard to decipher. DeepSeek says it took just two months and a little over $5 million to build one of its models, using degraded Nvidia chips that are available in China. I don’t believe that. Lennart Heim, who knows AI computing, pointed out that the cost numbers are almost certainly untrue, and the timing of this release, and the story of DeepSeek, is likely a strategic public relations campaign by the Chinese intelligence services.

More likely, the company had a large number of chips on which to train, export controls really only started biting in October of 2023, and the big new U.S. chip clusters haven’t really come online. Plus there are likely capabilities that U.S. firms are keeping private. A few open source Chinese models, though impressive, don’t mean the U.S. lead in AI is gone, or that the U.S. strategy of denying top chips to China has failed.

That said, no one denies that DeepSeek’s technical accomplishments are impressive, and the accomplishment suggest that there’s a lot about AI that we still don’t know. And a Chinese firm releasing a better cheaper product they claim was produced for less than most top AI project managers make in a year does make me feel like America is quite… Soviet.

More fundamentally, it reveals a number of truths about U.S. technological capacity.

The first is that Lina Khan was right when she pointed out that betting on monopolistic national champions, as we did with Boeing in the 1990s, is a disastrous national security strategy. Here’s what she said:

These days, the “national champions” argument often gets made in the context of our dominant tech firms. We often hear that pursuing antitrust cases against or regulating these firms will weaken American innovation and cede the global stage to China. These conversations often assume a Cold War-like arms race, with each country’s firms in a zero-sum quest for dominance…

History and experience show that lumbering monopolies mired in red tape and bureaucratic inertia cannot deliver the breakthrough technological advancements that hungry startups tend to create. It is precisely these breakthroughs that have allowed America to harness cutting-edge technologies and have made our economy the envy of the world. 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.

The second is that our large technical firms – Meta, Microsoft, Google, Amazon, Apple – are slothful bureaucracies, and no longer very good at developing and deploying terrific technology. As former Googler Praveen Seshadri put it, “Google has 175,000+ capable and well-compensated employees who get very little done quarter over quarter, year over year.”

That shouldn’t be a surprise, anyone who has tried to use the search function on Microsoft Outlook should be puzzled there are claims such a company can create God out of an algorithm. More broadly, as with Boeing, Intel, General Electric, et al, the incentives in the American corporate world are to win by innovation around monopolization and extractive techniques, not by building more and better products. Killing the underlying operational capacity of one’s firm always looks good, as you generate more and more cash. Until, as with Boeing, it doesn’t.

The third is that Chinese policy is, at least in some sectors, far more aggressive in forcing competition than in America, and it’s paying off for them. The Chinese electric vehicle industry leads the world, in part because of subsidies, but also because they have lots of companies that fight to produce better quality products at lower prices. A few years ago, the Chinese cracked down on their big tech firms and financial sector, which many around the world thought was disastrous for their economy. While Xi Jinping is not a good economic steward in general, this particular choice has been terrific. ByteDance is constantly churning out excellent products like TikTok, and now DeepSeek has rivaled or exceeded U.S. AI models. (It’s notable that DeepSeek is a spinoff of a leading hedge fund in China.)

And the fourth is that the anti-monopolists in the U.S. were right about their strategy, but our anti-monopoly laws take too long. In 2021, the Federal Trade Commission refiled a case against Facebook on the company’s acquisition of Instagram and WhatsApp. As FTC official Holly Vedova put it, “Facebook lacked the business acumen and technical talent to survive the transition to mobile. After failing to compete with new innovators, Facebook illegally bought or buried them when their popularity became an existential threat.” That same lack of skill pervades the company today; it’s why Facebook renamed itself Meta after the “Metaverse,” which looks in retrospect quite silly. If Facebook had been broken up, there would be tremendous innovation in social networking. But that case hasn’t even gone to trial.

Finally, something is broken with the AI story. It may still revolutionize everything, though it may also just be a very expensive version of Clippy. But the competitive advantages aren’t what we thought.

And after the paywall, some insanely good quality trolling by bank regulator Rohit Chopra, a new paper on the scam behind high utility rates, and an Arizona Congressman attempts to legalize robots prescribing pharmaceuticals because everything is now stupid.