Anthropic's Nasdaq debut puts its safety-first model under investor pressure

The Verge examined how the Claude maker, now publicly listed, must balance its AI safety commitments against huge compute bills and the need to keep growing revenue.

· 2 min read

A hand pointing at a smartphone displaying currency quotes against a laptop
Photo: Marga Santoso / Unsplash

Anthropic, the AI lab behind the Claude models, has gone public on the Nasdaq, and The Verge says the listing raises a hard question: can the company keep prioritizing safety while answering to public investors? The outlet explored the issue in an episode of its Decoder podcast featuring senior AI reporter Hayden Field.

According to the discussion, Anthropic is the first pure AI lab to list. Shares were priced at $83 and climbed as high as $111 on the first day of trading. The Verge describes the company as now among the most valuable in the world, though it says the exact figures matter less than what comes next.

A central pressure is cost. The Verge says Anthropic's compute bills add up to about $900 billion over the coming years, so it must keep raising revenue. Field cited an estimate from Steve Rattner that Anthropic would need to grow revenue roughly 17-fold within ten years to match the returns of some tech competitors. She said the company is still a long way from the profits that investors usually want to see.

Field also pointed to Anthropic's unusual corporate structure. Governance experts she spoke to said shareholders cannot vote out executives and have limited influence, because power sits with the founders and a special trust. Even so, she said, the demand to make money and buy compute could cause some things to fall by the wayside, possibly including expensive safety efforts.

On pricing, the discussion noted that Microsoft CEO Satya Nadella said at the Microsoft Ignite conference that the company aims to push token prices down. Microsoft's approach is to make AI vendors interchangeable under Copilot and route each task to the cheapest suitable model. The hosts said this sits uneasily with Anthropic's need for stronger margins on each token.

Field said Anthropic has already faced a backlash from individual customers who felt costs were being passed on to them. The company issued a public statement and partly rolled back the changes. She said it is also trying to lock users in with tools that are hard to replace, and is counting on enterprise and government contracts to bring in more money.

The report argues that Anthropic's safety and trustworthiness branding is especially important for large business customers. That branding now sits in tension with the profit side of the business. Field added that some small and mid-sized firms are building internal benchmarks to decide which tasks need a costly Anthropic model and which can go to open-source or cheaper ones. One example she gave was a split of 10 percent, 40 percent and 50 percent across three tiers.

The conversation also touched on OpenAI. Field said it appears content to let Anthropic go first, and that it can now point to Anthropic's debut as a precedent. She said she had heard that OpenAI CEO Sam Altman once badly wanted to list ahead of Anthropic, but that a run of setbacks and negative press this year changed the calculation.

Related