Anthropic goes public on commercial success, but that doesn't prove Amodei is right about AI that outperforms humans at almost everything

By The Motley Fool (via The Globe and Mail) · October 19, 2026.

The piece argues that the prediction by Dario Amodei, CEO of Anthropic, "is on track to come true." At Davos in January 2025, Amodei said AI could surpass humans "at almost everything" within two or three years, that is, around 2027-2028. The article's starting point is Anthropic's planned IPO, at a valuation of around $2 trillion (the article's own figure).

According to the piece, Anthropic is already ahead of OpenAI on several fronts. It reportedly posted revenue of $16.8 billion in the third quarter, in an update to investors. It would also be profitable on an adjusted basis, while OpenAI is said to be deep in the red. The article adds that Anthropic would go public in November and that OpenAI is pushing its own IPO back to 2027 to keep raising private capital. The article attributes these figures to the company in general terms. We have not been able to verify them against a primary source. In addition, the text mixes "quarterly revenue" and "annualized revenue" (*run rate*, the annual pace extrapolated from a short period), and the two should not be confused.

The argument about AGI (artificial general intelligence, capable of matching or surpassing people at intellectual tasks) rests on statements by third parties. Jensen Huang, CEO of Nvidia, said in March that AGI had already arrived and repeated it in September, with the launch of OpenAI's GPT-6 Astra. According to the article, that model was trained on more than 100,000 Nvidia Grace Blackwell GPUs and solved 10 open problems in mathematics and computer science. Sam Altman, by contrast, says OpenAI is "almost" there and expects to get there by the end of the year. Elon Musk points to Grok 5, expected next year. In other words, three leaders give three different dates.

Our take is that the article conflates two separate questions. One is whether Anthropic is a good business. The other is whether AI already surpasses humans at almost everything. Revenue and profitability answer the first. The second is answered by capability evaluations, and there a large number of solved problems does not amount to surpassing humans "at almost everything." That is why Amodei's statement cannot be considered fulfilled on the strength of a revenue figure.

There is also a conflict of interest readers should keep in mind. The one claiming AGI has already arrived is the supplier of the GPUs used to train it. And the text itself is syndicated content, with ads for other investment products, which The Globe and Mail states it has not reviewed. It is a piece aimed at investors, not a technical assessment. In general, the definition of AGI is so flexible that each player adjusts it to its commercial calendar.

The most useful part of the story is what it says about the transition. The article itself acknowledges that "the biggest problems are not the technology, but what it means for society." It mentions political pushback against data centers and anxiety in the job market. That is the difficult short term that shouldn't be glossed over: employment disruptions and the concentration of power in a few companies that are also publicly traded.

In the long term we still see a better horizon, with medicine that eradicates diseases and abundance that frees people to work on what they are passionate about. A lab making money helps fund that path, but it doesn't guarantee it. In the meantime, the practical rule is to keep three things separate: what benchmarks measure, what CEOs say and what the stock market values. Only the first describes actual capability.