Something special is going to happen in 2026: both OpenAI and Anthropic are going public. This will finally mean that individual investors can invest in them and bet on their future with their money. It will be the definitive test for the credibility of companies that have grown exceptionally in recent years but that have also burned money as if there were no tomorrow. But be careful, because there is a compelling reality here: they are going to continue burning it in an even more astonishing way.
The two sides of the IPO. The Wall Street Journal has had access to the financial documents presented to investors before the IPOs proposed by both OpenAI and Anthropic. They reveal extraordinarily striking data that have two sides.
Amazement and concern with OpenAI. For example, OpenAI has indicated that it will almost double its revenue this year. According to their forecasts, they could become profitable in 2026 if one excludes the cost of training their models (which are stratospheric, of course). But there is the other reality: OpenAI expects to spend $121 billion on computing power in 2028, so even doubling revenue it will lose, attention, $85 billion. No company has ever lost this amount of money and survived, but OpenAI not only promises that it will survive, but that those losses will end up being almost anecdotal.
I tell you the truth, but only part of it. Both companies wanted to show two different versions of reality when talking about how they present their profitability. In one, the very expensive model training processes are included, and in others in which these costs are excluded under a heading called “computing for research.” Excluding those costs, OpenAI is on track to achieve a small pre-tax operating profit this year. Anthropic also promises to achieve this if its most optimistic scenario comes true.
Excluding the cost of training models, both OpenAI and Anthropic could be “profitable” this year. Source: WSJ.
Until 2030, no real profitability. If the costs and investment in model training are included, OpenAI indicates that it will end up being profitable in 2030, a fact that they had already predicted a long time ago and that could not hide a conclusive reality: not only has the company not stopped spending money until now: it is going to continue spending it, but to an even greater extent with projects like Stargate at the helm. Saying that in 2026 they will be profitable if we do not consider training costs is like an airline telling us that it is profitable excluding the cost of fuel. Anthropic, by the way, expects to be fully profitable in 2028.
Revenues growing fast, costs even faster. In addition to these training processes, both OpenAI and Anthropic are spending billions of dollars each year on inference, a section that is beginning to be even more important at an operational and strategic level. Currently, these inference costs represent half of each company’s income, although it is expected that inference technology will become cheaper and therefore the costs as well. Here, however, there are two big differences between both companies:
- OpenAI: most ChatGPT users do not pay to use the service, so OpenAI assumes these inference costs without making them profitable. According to OpenAI, this facilitates adoption and will allow users to become subscribers in the future, something that is not happening too much at the moment.
- Anthropic: This startup has managed to win over many companies that pay to use their models, and it is evident that the company is absolutely focused on making you pay to use their models if you want to use them. And if not, tell OpenClaw.
Betting on the future. The companies and venture capital funds that have invested billions in OpenAI or Anthropic have made a bet on the future. They have blind faith that these companies will end up taking over the world, so the fact that they are still unprofitable today does not scare them… or not enough to withdraw from this expensive race. Both have experienced spectacular growth that serves as an argument for investors. In addition, the growing interest of companies in integrating AI solutions by paying for them has boosted Anthropic and even caused OpenAI to reorganize and change its strategy. Less fireworks and hypemore focus on what makes money.
The IPO as a trick to survive. Both companies are going to continue burning money like there was no tomorrow in the coming years, but now they hope that investors will be the ones to sustain their businesses. The amount of money they will need has even made the Nasdaq make things easier: it will allow newly listed companies to join its renowned index more quickly, which would give them access to larger capital reserves. Now it will be the public market and to a large extent the individual investor who will decide whether they want to bet on that future or not.
A small survey. Would you invest in OpenAI or Anthropic if it went public? It is evident that both companies generate different impressions, and although their strategies and ways of doing things are different, it is clear that this public sale offer is going to be very striking when it occurs. So, it is a good time to find out a little about what you, the xatakeros, think about this financial movement of these companies.
Image | TechCrunch | Wikimedia Commons
In WorldOfSoftware | NVIDIA has so much money that it is becoming something different: the largest startup incubator in the world
