# Anthropic's public listing puts its safety mission and governance under scrutiny

published: 2026-11-22T06:05:33.923Z
topic: infrastructure
source: https://www.theverge.com/podcast/1038701/anthropic-ipo-ai-safety-openai-nvidia-structure
url: https://aiagentnews.io/stories/anthropic-s-public-listing-puts-its-safety-mission-and-governance-under-scrutiny

Anthropic, the AI lab behind Claude, has gone public on the Nasdaq, and Verge hosts discuss how public-market pressure could affect its safety commitments. The conversation focuses on the company's large compute costs and a governance structure in which its founders hold most of the voting power.

Anthropic went public on the Nasdaq, and Verge senior AI reporter Hayden Field and host Nilay Patel say the stock was priced at $83 a share and rose as high as $111. Patel puts the company's expected compute bills at roughly $900 billion over the coming years, and the two argue that revenue will need to grow quickly to cover them.

On governance, Patel says Founder LLC holds a single share with 50.1 percent of voting power, and its seven co-founders own it. Field says the Long-Term Benefit Trust, a four-member group that includes former Federal Reserve Chairman Ben Bernanke, can appoint or remove board members but lacks authority over everyday decisions such as model releases.

The pair also say Anthropic's safety branding may clash with its margins, and that large enterprise customers could demand data portability that weakens the company's lock-in. Field reports that small and mid-sized businesses are building internal tools to send simpler tasks to cheaper or open-source models.

## Why it matters if you build agents

If you build on Claude, expect pricing and model-choice pressure as Anthropic answers to public shareholders. Keep evaluation tests for your own tasks so you can move work between models without rewriting your agent.

The speakers expect many customers to split workloads across frontier, mid-tier and open-source models, so routing logic is becoming a standard part of agent design.
