Anthropic's IPO shrinks from $2 trillion to ~$1.1-1.3 trillion: the listing becomes a referendum on all of AI

By Axios · November 7, 2026.

The valuation Anthropic is seeking for its IPO has plunged in a month. According to Axios, the company entered October aiming for $2 trillion. By mid-month, investors, wary of slowing revenue growth in the third quarter, were already talking about roughly $1.5 trillion. After Thursday's market rout, the Wall Street Journal put the talks at around $1 trillion and Bloomberg at between $1.1 trillion and $1.3 trillion. These are figures reported by the press based on investor sources, not an official price.

The market backdrop explains much of the shift. The Nasdaq Composite closed Thursday down 5.5%, now in correction territory and nearly 12% below its October 6 high. The SOX semiconductor index lost 12.3%. Axios links the sell-off to fears of a slowdown in data center construction following Democratic gains in the midterm elections. On Friday, the jitters eased only partially.

Anthropic matters beyond itself for a specific reason: it carries some $900 billion in compute and infrastructure commitments with Amazon, Google, Microsoft and Broadcom, among others. Its financial health ripples through to chipmakers, networking companies, data center operators and energy providers. If its valuation suffers, it calls into question the demand assumption on which that entire chain has been built.

Max Kettner, chief multi-asset strategist at HSBC, described to Bloomberg Radio a market driven by narratives. The dominant one now, which he does not share, would be that the AI cycle is dead. His prescription: a conservative valuation, even below $1 trillion, to keep the IPO from repeating the SpaceX case, whose shares reversed their initial euphoria and traded below the offering price for more than a month.

Our take: the markdown is a price correction, not a verdict on the technology. Going from $2 trillion to a range of $1 trillion to $1.3 trillion still leaves Anthropic as one of the most valuable companies on the planet; what has changed is the euphoria premium. That the cited reason is a slowdown in the pace of revenue growth, not a decline, is an important distinction: growing less quickly from an enormous base does not mean the product does not work.

Even so, there is a structural risk worth naming. Committing close to $900 billion to infrastructure before revenue can back it is a long-term bet financed by short-term expectations. When expectations reset, the bill falls on the entire chain: suppliers, data centers and, eventually, the jobs tied to that build-out. It is the kind of hard transition that should not be glossed over.

And yet, the underlying logic does not hinge on the quarter. Valuation bubbles and useful infrastructure often coexist: the excess capacity of one phase tends to make compute cheaper for the next, and cheap compute is the raw material for what really matters in the long run, from drug discovery to the abundance we advocate. A prudent IPO, priced to leave room for upside, would be better for that horizon than a euphoric debut followed by disappointment.

What to watch: the date (the article says it could happen later in November, without confirmation), the final price versus the leaked range, and the stock's performance in its first weeks. That last data point will say more about the market's mood toward AI than any headline beforehand.