Anthropic’s IPO chatter and pricing shift
Social monitoring shows strong pre‑IPO momentum for Anthropic — Polymarket odds put an IPO before OpenAI at 64% and pre‑IPO share demand is reportedly up about 50% compared with OpenAI’s -10% signal (x.com) (x.com). The company is also rolling out usage‑based pricing to address compute constraints while investors debate big valuation offers and enterprise revenue claims in the market chatter (x.com) (x.com).
Anthropic is drawing fresh public-market chatter as traders on Polymarket now give it roughly a 62% chance of going public before OpenAI. (polymarket.com) That bet comes as Anthropic is pricing its models more directly by usage: Claude Opus 4.7 costs $5 per million input tokens and $25 per million output tokens, the same as Opus 4.6, and Anthropic says developers can cut costs with prompt caching and batch processing. (anthropic.com 1) (anthropic.com 2) In plain terms, usage-based pricing means customers pay for how much text they send in and get back, not a flat software seat. Anthropic has also added spend controls for business plans, letting administrators set organization and user limits as companies try to manage fast-rising model bills. (anthropic.com 1) (anthropic.com 2) The backdrop is a company that raised $30 billion on February 12, 2026, at a $380 billion post-money valuation. Anthropic said that round would fund research, product development, and infrastructure, and its chief financial officer, Krishna Rao, said customer demand was driving the financing. (anthropic.com) Anthropic also used that February funding announcement to make unusually detailed operating claims: a $14 billion run-rate revenue figure, more than 500 customers spending over $1 million annually, and eight of the Fortune 10 as Claude customers. Those numbers have become central to investor debates over how quickly the company could justify a public listing. (anthropic.com) The IPO talk is still talk. Polymarket’s contract resolves only if Anthropic actually completes an initial public offering before OpenAI by December 31, 2027, and a separate Polymarket market recently showed traders still assigning very high odds to no Anthropic IPO by June 30, 2026. (polymarket.com 1) (polymarket.com 2) Anthropic has not published an S-1 filing or announced IPO terms on its newsroom page. Instead, the company has kept rolling out products, including Claude Opus 4.7 on April 16, 2026, and expanding governance ahead of what would be heavier public scrutiny. (anthropic.com) (anthropic.com) One governance move landed this week: Anthropic said on April 14 that Novartis chief executive Vas Narasimhan joined its board, and that directors appointed by its Long-Term Benefit Trust now make up a majority of the board. Anthropic said the trust exists to balance stockholder interests with its public-benefit mission. (anthropic.com) That leaves Anthropic in a familiar late-private-company position: huge valuation, fast revenue claims, expensive infrastructure needs, and a pricing model built around rationing and monetizing scarce computing power. The next hard signal is not social chatter or prediction-market volume, but whether Anthropic files to sell stock to the public. (anthropic.com) (polymarket.com)
