Anthropic cuts its IPO target to $1.5 trillion: the public market will decide what the compute bet is worth

By Benzinga · November 8, 2026.

Anthropic is pressing ahead with its initial public offering (IPO) this year despite jitters across the sector. According to reports cited by Benzinga, the company has reportedly lowered its valuation target to $1.5 trillion, down from $2 trillion previously. It aims to raise around $100 billion. That is a 25% cut to the target, and the fundraising figure remains enormous.

Benzinga provides company figures for the quarter. Third-quarter revenue was $18.6 billion, and Anthropic ended the period with more than $95.6 billion in cash. But two customers account for a fifth of its revenue. And the company expects to spend around $895 billion on capital expenditure over the coming years.

The article reviews how other recent IPOs have traded. According to Benzinga, SpaceX rose as high as $225 and fell to $104 within a few weeks. Circle went from $60 to around $300 and then hit a low of $50 in February. Cerebras, the AI chipmaker, more than doubled its price on its first day, reached $386 and now trades around $89. It is the classic pattern: initial euphoria, profit-taking and a fading narrative. It is a historical pattern, not a law, and the article itself presents it as a clue, not a forecast.

The piece adds regulatory and competitive risks. According to Benzinga, Democrats won the House of Representatives and the Senate on a promise to regulate AI. Recent security incidents in which AI has attacked companies have intensified that pressure. At the same time, low-cost platforms such as Kimi and DeepSeek are squeezing from below, while OpenAI, Grok and Gemini compete with deep pockets.

**Our take.** The lowered target matters less as a sign of weakness than as a price adjustment. A lab that brings in $18.6 billion in a quarter remains an exceptional growth business. But that business is financed by capital commitments of unprecedented rigidity. The IPO will be the first time public markets put a price on that tension between growing revenue and spending on a never-before-seen scale.

We see three weak points. First, concentration: if two customers make up 20% of revenue, the growth story hinges on a handful of renewals. Second, pricing pressure: when Kimi or DeepSeek offer near-comparable quality at much lower cost, closed-model margins narrow, and a listing demands margins. Third, regulation. We advocate evidence-based governance, but the market will price it in as uncertainty even if it is sensible.

In general, those who sell the infrastructure (chips, data centers, energy) get paid sooner and with less risk than the lab that buys it. That asymmetry will weigh most heavily on results in the coming years.

In the long run, our thesis is unchanged. Investment of this magnitude in compute is the clearest signal that the sector is betting on AI capable of accelerating medicine and science, and of making nearly everything cheaper to produce. That promise, of eradicating disease and generating abundance, requires patient capital. But the path will not be linear: the stock market is likely to punish in the short term what history rewards in the long term. A stellar debut followed by a correction would not be a failure of the project, but the market discovering the price of a transition that remains hard to value.

For readers, the lesson is to separate three things: product quality, share price and the pace of spending. All three can move in different directions for months.