Anthropic heads for the stock market and tests the price of AI

On June 2, 2026, Superhuman reported that Anthropic had confidentially filed a draft S-1 registration document for a stock market listing. The newsletter wrote that the company’s valuation could move toward one trillion dollars. The same item mentioned that Anthropic is expanding access to its Mythos model for a European cybersecurity agency. If confirmed in official listing documents, this would be one of the first major tests of how public markets value companies built on expensive computing power. The listing itself is not the only important point. An IPO would show investors more detailed figures for the first time: revenue, operating margins, infrastructure costs and obligations to suppliers of chips and data centres. Between 2024 and 2026, private capital often valued AI companies according to the size of the expected future market. The stock market will want to see whether fast growth can turn into profit. Superhuman adds a broader picture of the startup market. According to the newsletter, more than 250 billion dollars of venture capital flowed in 2026 into just two firms, OpenAI and Anthropic. It also cites PitchBook data saying that almost half of 857 US unicorns had not raised new funding for three years and that more than 220 former billion-dollar startups had fallen below that level. AI is not only attracting capital. It is pulling it away from other parts of the technology market. The phrase unicorn graveyard in Superhuman’s text captures the problem: capital has not disappeared evenly, but has concentrated in a few names. That helps the biggest AI firms pay for computing capacity, hire researchers and keep pace in model development. Smaller startups without a direct AI link or clear growth have a weaker negotiating position. For European companies and regulators, the reference to Mythos also matters. If Anthropic provides access to a security model for a European cybersecurity agency, it shows AI moving from ordinary productivity into sensitive infrastructure. Cybersecurity is an area where a powerful model can find vulnerabilities, but can also help attackers. Access will therefore probably be controlled, contractual and audited. For Czech managers, the lesson is practical: when choosing AI suppliers, functionality is not enough. They must also watch financial stability, dependence on computing power and data rules. If AI becomes a core layer of company processes, a supplier failure or sharp price increase will not be a technical inconvenience. It will be an operational risk. Terms: S-1 is a registration document for a company entering the US stock market. IPO means the first public offering of a company’s shares to investors. A unicorn is a private startup valued at least at one billion dollars.

Anthropic heads for the stock market and tests the price of AI

The full article is available to users registered at Hard Skills