Sam Altman released ChatGPT for free in late 2022. Many users fell in love both with it and Altman’s argument that artificial general intelligence could be achieved through scaling. The bottleneck became compute and building data centers the goal of the global economy.

Altman has largely achieved his goals. Data centers have become the largest target of capital investment in the history of humankind, larger than the internet and railroads, most economic growth is now from data centers, and many of those data centers serve OpenAI. For instance, 70% of the demand for Microsoft’s cloud centers comes from OpenAI.

Users fell for Altman’s argument because, on the surface, ChatGPT and the underlying technology, large language models (LLMs), worked well. Their seemingly perfect prose seemed so intelligent, so human-like that it was easy to think that LLMs would replace everyone, particularly as superintelligence was achieved.

It was particularly “easy to understand for most non-technologists — a chief executive, board member or investor — a healthy portion of whom began demanding their organizations immediately embed the technology throughout their operations,” according to the former head of IT for Lululemon, an apparel maker.

And those people are still enamoured. Numerous surveys have found that workers are far less optimistic about AI than are top managers, and board members are the most optimistic. In other words, the further you are from real work, the more optimistic you are about AI.

These easy to impress users touted the ease of writing emails or reports, summarizing them, building websites, and later creating images and videos, and writing code. They ignored the hallucinations and slop because that was for downstream workers to handle, and presumably they would disappear as AGI was achieved. The non-technologists were now in charge of AI.

Elites also jumped on the bandwagon expecting huge productivity benefits. Professors, heads of big consulting companies, politicians, and policy makers were some of those who made optimistic predictions while many low-level tech people remained less effusive, often silent and sometimes critical.

The Results for Businesses

From a business standpoint, the technical people were mostly right. Generative AI has largely been a failure, both for users and for suppliers. Published last fall, an MIT Study found that 95% of AI initiatives fail to achieve a positive ROI. A subsequent study from software company Atlassian reported similar levels of failure. Some attribute these failures and worker resistance to AI slop or, according to Cory Doctorow, author of “Enshittification: Why Everything Suddenly Got Worse and What to Do about It,” and most recently “The Reverse Centaur’s Guide to Life After AI,” takes this argument further. He reminds us that “workers actually wanted earlier technological breakthroughs and often had to fight to get them into the workplace,” unlike AI, which is pushed by “non-technologists.” “With AI, people are more likely to feel that the technology is being shoved down our throats; some workers are even required to use it.”

“If you look back to the business press of the 1980s and late ’90s, it’s full of hand-wringing editorials about how bosses will cope with workers who are smuggling in the web. You look at those same press outlets today, and it’s full of people saying, “What are we going to do about the fact that no one in the workplace wants to use AI?”—along with ads for firms that will spy on your workers for you so that you can punish the workers who refuse to use AI.

The Results for AI Labs

The business results for AI Labs are also currently poor even if we ignore the low-cost models from China. Losses for OpenAI are more than twice their revenues because the prices were set far too low, encouraging usage but meaning it requires constant investments and loans, often from other members of the value chain such as hyperscalers (Google, Amazon, Oracle, Microsoft) and semiconductor suppliers (Nvidia, AMD, Broadcom, TSMC, Samsung). Those companies have covered OpenAI’s losses, first with investments and later with loans through circular financing and the creation of intermediary companies.

One result is that although hyperscalers once had huge positive cash flows, now those cash flows are about to turn negative from their close to $1 trillion spending on data centers each year. And that spending is not about to stop. According to Moody’s, the five largest hyperscalers now have $662 billion in lease obligations for data centers.

SoftBank is probably the closest to bankruptcy. It has now committed $65 billion to OpenAI and banks now require it to pay nearly triple (7.88%) what’s ordinarily required because Softbank wanted to use OpenAI’s valuation as collateral. The stock market has also been punishing Softbank, first pushing share prices down 40% from October 2025 and later 1/3 from this June following a rally in early 2026.

The types of financial risks faced by Softbank and others exposed to OpenAI and Anthropic are generally ignored because some users are very happy with AI, and because the tech bros continue to make “extraordinary claims without extraordinary evidence” almost weekly. AGI (artificial general intelligence) is imminent, as are big layoffs, cures for cancer, big scientific advances, and 1,000 times improvements over the next six years.

Along with the extraordinary claims there are also tricks, some of which constitute fraud. Hyperscalers are hiding $3 trillion of debt in their balance sheets as they spend hundreds of billions on data centers, and are paying higher interest rates than others. They are also reporting the rising valuations of OpenAI and Anthropic as income, even though the valuations are partly determined by them.

The goal of the tech bros

The goal of the tech bros is to push the IPOs of OpenAI and Anthropic onto retail investors just as they did with SpaceX and before that with a long list of money-losing Unicorns, which are detailed in my book, Unicorns, Hype and Bubbles.

Whereas 23 companies founded between 1975 and 2005 were profitable and in the top 100 for market capitalization for at least two years, only one company founded since 2005 (Uber) has achieved the same, but venture capitalists have made money by convincing retail investors to buy shares of their companies.

The tech bros aren’t concerned with whether AI startups ever become profitable, they want to salvage their huge investments. They are paying an army of influencers to push this year’s IPOs for OpenAI and Anthropic just as they did with SpaceX. Retail investors and the public in general can escape this trap by not investing in their IPOs, thus preventing them from stealing trillions. In my opinion, avoid investment in OpenAI or Anthropic IPOs because you could lose a lot. Potentially, only the earliest investors will win.