Anthropic Says Its Market Is $30 Trillion, Same As US GDP

Anthropic is apparently telling investors its potential revenue could reach $30 trillion, about the same as US GDP ($32.5 trillion). The number is misleading because, as The Wall Street Journal points out, it’s a “total addressable market,” or the potential revenue for the entire sector. As the paper notes, the $30 trillion figure assumes Anthropic will capture 100% of the market.

The Journal reports, “While they have always involved a bit of guesswork, these estimates are especially squishy when it comes to trying to predict how the rapid adoption of artificial intelligence will upend entire industries.” Nevertheless, the figure is wild at a level which is impossible to express.

Leave aside the figure, which has been passed around parts of the AI and investing world that have to raise money when Anthropic goes public. Analysts put Anthropic’s revenue this year as high as $65 billion. Its second quarter revenue was $11.6 billion. At its current growth rate, it would not be a long shot to reach $200 billion in 2027. That is about the same as Meta’s (NASDAQ: META | META Price Prediction) was last year.

Then there is a list of risks, which is well articulated, and some are plausible. At the top of that list is the risk of AI data centers. Depending on who is doing the forecasting, total annual investment in these will reach the trillions of dollars by the end of the decade.

Among the data center risks is that they will be either too big or too small. If the AI explosion continues, infrastructure, electricity, chips, and land may fall short of meeting colossal demand. The human factor is that residents in areas where these might be built, and politicians, have started closing off the locations most desirable to builders. It’s like a plane running out of runway.

At the far end of the spectrum, data centers could be overbuilt. AI adoption, partially among enterprise users, could slow quickly. AI could simply be too expensive based on its intelligence and efficiency results. Chinese models could be cheap enough to undercut demand for more expensive products from OpenAI and Anthropic. Revenue forecasts for these companies could be significantly overblown. No one knows because there is no data on what the world of AI will look like in a year, two years, or five.

One data center challenge is where the money will come from. Mega-tech public companies have run out of money. Financial firms have stepped in to trade potential huge profits weighed against a collapse. As the financial system goes, big financial players sell some of their risk to other institutions, or even the public, via publicly traded instruments. It may well be a house of cards.

Part of the recent analysis of data centers is that insurance companies have started to shy away from covering them entirely. One factor apparently considered very risky is terrorism. Since this is often mentioned as a risk to the US grid, why exclude data centers?

The financial risk is really impossible to anticipate. If AI kills millions of jobs, particularly in America, and unemployment rises to 10% or even 20%. What happens to the tax base? What happens to the customer bases of many companies that rely on consumers? AI will have eaten itself alive.

The idea that Anthropic can have $30 trillion in revenue, or even a large fraction of that, assumes a far-end-of-the-spectrum AI success.

The Wall Street Journal headline reads “Anthropic Expected to Tell Investors It Sees Over $30 Trillion in Potential Revenue.” If so, AI can start its own nation.

Contact [email protected] for any questions or corrections.