Anthropic’s $2 Trillion IPO Includes $120.5 Billion Compute Plan

By Nathan Brooks

Anthropic has filed for an initial public offering that could value the Claude maker at about $2 trillion, according to its Form S-1 registration statement with the U.S. Securities and Exchange Commission. The filing would place the company among the technology industry's largest potential stock-market debuts.

Anthropic plans to list Class A common stock on the Nasdaq Global Select Market under ticker “ANT.” The company has yet to determine the number of shares or the price range.

Reuters reported that the offering could raise as much as $100 billion. Anthropic has not confirmed that amount as an offering term, so the fundraising figure remains a reported possibility rather than a filed target.

What do the operating numbers show?

Revenue reached $16.3 billion in the first half of 2026, compared with approximately $1.2 billion in the same period last year. Anthropic also reported improved operating performance, although its bottom line remained deeply negative.

The reported net loss reached $80.7 billion. Accounting adjustments related to financing instruments contributed substantially, so investors should not interpret the figure as $80.7 billion in cash spent running the business.

Anthropic warns that its revenue growth rate may not be sustainable and could decline. The company may also need additional financing sooner than expected.

It does not anticipate paying dividends in the foreseeable future. The expansion still depends on access to increasingly expensive computing resources, placing compute procurement at the center of the offering.

Revenue acceleration helps. The loss and infrastructure obligations complicate the arithmetic.

How large are the compute commitments?

Anthropic expects to spend up to $120.5 billion on compute involving SpaceXAI and Nvidia. It also expects up to $114.5 billion with Google through 2033.

Those two stated spending limits add to $235 billion. The filing gives no breakdown by product, unit volume, delivery timing or annual schedule.

For Nvidia, the clearest semiconductor demand implication could be accelerator procurement. Yet Anthropic names no product family, process node, wafer size or shipment schedule, so the commitment supports direction rather than a precise chip forecast.

Bloomberg Intelligence technology research head Mandeep Singh viewed improving compute efficiency and higher gross margins as positive signs. “Gross margins are inching closer to 50%, whereas just a quarter back… it was like… sub 20%,” he said.

Singh's margin observation is an analyst assessment of disclosed performance, not a formal company forecast. Even so, approaching 50% would not erase the scale of spending required to train and run advanced models.

Why do the safety warnings matter?

Anthropic's S-1 warns that advanced AI could create what it calls catastrophic or existential risks to humanity. The filing cites dangerous capabilities, deceptive behavior and systems that become difficult for humans to control.

“Our development of highly advanced models, platforms, and applications and expansion of use cases could further increase the risk that our models cause harm,” Anthropic stated in the prospectus. The warning directly links broader deployment with greater potential for harm.

The company says its safeguards include pre-release testing, red-team assessments and systems designed to detect harmful outputs. It may delay training or restrict model releases when safety measures prove insufficient.

It also acknowledges that its safeguards may not succeed. Singh characterized the worst-case scenarios as typical risk disclosure for emerging technology companies preparing to go public, where such wording can help address potential liabilities.

The legal framing is familiar; the technical language is blunt. For investors, safety controls become a business constraint rather than a separate policy discussion, because delays or restrictions can affect model releases and use-case expansion.

Who moves first?

OpenAI has also prepared for a potential IPO, working to scale revenue and computing capacity. CEO Sam Altman said in September that the company will not go public in 2026, citing AI safety concerns and saying it has more work to do before pursuing a listing.

Cerebras, which develops wafer-scale processors for AI workloads, completed its Nasdaq IPO in May after filing for a second time. Its completed listing contrasts with Anthropic's still-unconfirmed timetable.

Anthropic reportedly targets a mid-November IPO, with November 19 or 20 floated as possible listing dates. The company has confirmed neither those dates nor a broader timetable.

If the offering proceeds, pricing will show how public markets value Anthropic's AI revenue against its infrastructure obligations. Margin progress toward 50% and delivery on the compute plan will determine whether growth can outrun those commitments.

The practical test is whether revenue and margins keep pace before additional financing becomes urgent. If Anthropic proceeds in mid-November, its offer price will reveal how investors value that tension.