Anthropic's $1.4 Trillion IPO Is a Referendum on AI Itself

Anthropic is seeking a valuation of up to $1.4 trillion and plans to raise $90 billion in its initial public offering, according to its recent regulatory filing. The AI startup's road show began yesterday, with shares expected to price between $70 and $78. No official go-public date has been set, but the end of next week appears the likely candidate.

The filing's headline numbers describe a best-case scenario, not a guarantee. It is now up to Anthropic and its bankers to convert that scenario into demand among public investors. One variable is already locked down: backers have committed to buying $26 billion in Anthropic shares at the IPO price, whoever the wider market decides to be.

Those committed backers include Broadcom, Google, and Nvidia — some of the biggest names in the AI race itself. The arrangement feeds an ongoing criticism of the industry: that it relies too heavily on circular financing, with AI's largest suppliers and partners also serving as its anchor buyers. Anthropic has not said publicly how it would respond to that critique.

Joe Ciolli, author of Business Insider's First Trade newsletter, unpacked the decision to pair an AI-heavy private placement with the IPO. His framing is blunt: Anthropic's IPO is an investor referendum on AI.

What does the referendum actually decide?

The offering gives public investors a chance to back — or avoid — a new bellwether for the entire AI industry. The last time the general public weighed in on AI, things did not go well for the technology, as the midterms showed.

Investors, of course, are not general voters. Their concern is less about data centers in their backyard and more about whether Anthropic is a business that will keep the stock price rising after they buy in. The company's own IPO paperwork complicates that pitch: it shows the soaring compute costs required just to keep revenue growing. A growth story that depends on ever-larger spending on chips and infrastructure is, on its face, not an easy sell.

Is there a precedent for a stumble?

Brian Boland knows what a difficult public debut looks like. A vice president at Facebook during its disastrous 2012 IPO, he spoke to BI's Stephen Council about that experience and the similarities he sees with Anthropic's looming listing.

Facebook — now Meta — eventually figured it out. But recovery took more than a year before the stock closed above its IPO price. With so much riding on AI today, the stakes run far higher than they did for a single social network in 2012. It remains unclear whether the market could stomach an AI darling underperforming for that long.

Who wins regardless of the share price?

One group is already coming out ahead, no matter how the stock trades — besides early investors and employees. Anthropic has budgeted a staggering $41.5 million in legal fees for its IPO. That figure runs nearly double what SpaceX planned to spend during its record-breaking flotation.

For context, that $41.5 million is pure transaction cost — money paid to lawyers before a single share changes hands. It signals both the complexity of Anthropic's corporate structure and the scrutiny the listing will attract.

The pricing range of $70 to $78 gives Anthropic a limited band in which to negotiate demand. A best-case $1.4 trillion valuation assumes the road show lands near the top of expectations. If investors balk at the compute-cost math in the filing, the deal's final terms could land well below the filing's ceiling.

Anthropic's debut, whenever it lands, will be read as a verdict on AI economics. A strong pricing validates the model of spending vast sums on compute to sustain revenue growth; a weak one revives the circular-financing critique with public-market evidence behind it.