Anthropic is weighing a plan to set aside between 15% and 20% of the shares in its forthcoming initial public offering for retail investors, according to people familiar with the matter. JPMorgan is expected to lead the US retail portion of the offering. The company’s S-1 prospectus does not confirm a specific retail allocation; it says only that the banks may reserve shares for their online brokerage customers. The reported plan could not be independently verified.

The move would follow a broader push by large technology listings to court individual investors. SpaceX reserved about 20% of its June listing for retail buyers, after initially planning 30%, according to reports. Banks have increasingly used retail demand as a way to broaden the investor base and, in some cases, support pricing. One analyst said individual money is less sensitive to price, and a book that leans on it lets banks price a deal higher than institutions alone would support.

Warren sets a deadline

Senator Elizabeth Warren has sent Anthropic chief executive Dario Amodei a four-page letter demanding answers by 3 November about the company’s IPO filing. The Massachusetts Democrat asked how Anthropic can ensure its technology will not become an existential threat to humanity. She also asked about the company’s ties to Amazon and Broadcom and how it will justify a valuation that could approach $2 trillion. Anthropic did not immediately respond.

Warren wrote that the risks 'demand greater scrutiny before they become intertwined with the public capital markets,' according to a copy of the letter. The intervention adds a political dimension to what is already one of the most closely watched AI listings. It also puts the company’s safety claims and its corporate relationships under a public lens at the moment when investors are being asked to price those claims.

Safety claims meet public markets

Anthropic was founded in 2021 by former OpenAI executives and researchers and has built the Claude family of AI models. It has positioned itself around safety and reliability, arguing that advanced systems need strong safeguards. It is also a public benefit corporation, a structure that lets it pursue a mission alongside profit. Going public would test how those commitments survive the demands of quarterly reporting and shareholder returns.

The filing acknowledges concerns inside and outside Anthropic about where advanced AI could lead. It warns that sufficiently advanced systems 'could, if not developed, deployed, and operated with appropriate safeguards, oversight, and security measures, pose catastrophic or existential risks to humanity.' It also notes that Anthropic could become a political target.

The document says proposed US laws contemplate the nationalisation or mandatory shutdown of frontier AI systems. It adds that the rules Anthropic itself argues for could slow its growth. Government contracts make up about 1% of revenue, so direct public-sector exposure is limited, but the regulatory risk is not.

The same document sets out the size of the opportunity. Anthropic says knowledge work was worth more than $40 trillion globally in 2025, a figure that tops SpaceX’s $28.5 trillion market estimate. That comparison is meant to show how much economic activity the company believes AI could reorganise. It also helps explain why investors are willing to consider a valuation near $2 trillion even as the company warns about catastrophic risk.

Compute commitments pile up

Anthropic has signed a series of large compute agreements with major cloud and chip partners. The company has deals with Google, Amazon, Microsoft and Broadcom. It also agreed to spend up to $120.5 billion with SpaceXAI, the SpaceX unit formerly known as xAI. That works out to about $3.1 billion a month from December. After early 2027, either side can cancel on 90 days’ notice.

The company agreed to spend up to $44.6 billion with Nscale and about $41.3 billion with Lambda. It expects to pay $125 billion to $150 billion for more TPU chip capacity from 2028 to 2031, outside its commitments table. Since 30 June, it has signed deals with about $245 billion in minimum payments. The total is still blank. Reports have said Anthropic could spend $750 billion on compute.

Those numbers matter because compute is both the engine and the cost base of the business. The company’s ability to serve customers depends on securing enough chips and data-centre capacity. But the obligations also create fixed costs that must be met even if revenue growth slows. The filing’s disclosure that the total commitment figure is blank adds to the uncertainty around the full scale of the company’s future spending.

Margins and who pays

Cost of revenue, mostly compute that serves customers, was $2.38 billion in the first half, against $16.3 billion of revenue. That implies a gross margin of about 85%. The figure leaves out training. Total compute spending was $8.95 billion.

The filing cautions that revenue growth may not continue at comparable rates. 'We do not expect our revenue will continue to grow at comparable rates, and we cannot assure you that it will continue to grow at all,' it says. Margins may also shrink as new compute arrives before the revenue it supports. Models released in the six months to 30 September made up over 80% of third-quarter revenue.

Sales through Amazon, Google and Microsoft’s cloud marketplaces were 48% of first-half revenue, up from 11% in 2023. Google has an option to buy up to $10 billion of stock at the IPO price. Anthropic values the convertible notes Amazon holds at $108.6 billion; Amazon puts them at $97.9 billion. The difference highlights how much of the company’s capital structure depends on valuation assumptions and on the cloud partners that also sell its products.

Retail demand and the valuation test

A retail allocation of 15% to 20% would be large for a technology IPO of this size. Retail investors are often given a smaller slice, with most shares going to institutional funds. A bigger retail book can widen the shareholder base and generate public attention, but it can also bring more trading volatility after the first day. Banks must balance those risks against the benefit of extra demand.

At a valuation near $2 trillion, Anthropic would rank among the most valuable companies in the world. Its first-half revenue of $16.3 billion annualises to roughly $32.6 billion, which would still imply a rich multiple if growth slows. The company’s argument rests on the size of knowledge work, the spread of AI into white-collar tasks, and its position as a leading model developer. The counterargument is that compute costs, regulation and competition could compress margins before those ambitions are realised.

Some shares can be sold from the first trading day. Co-founders and holders of more than 0.5% must wait until after third-quarter 2027 results, or 365 days. Most other holders are free after first-quarter 2027 results, or 180 days. The lockup schedule will shape the supply of stock after the listing and could influence how investors approach the first months of trading.

A critic wrote that the company described in the S-1 'no longer exists,' pointing to the blank third-quarter figures. Another research note called Anthropic 'a much, much better business' than it had expected, but said it would not yet buy. The split captures the debate: the company has shown rapid revenue growth and strong gross margins, but the filing also leaves important figures incomplete and asks investors to accept a valuation based on a future market that is still being built.

The first trading day will test whether retail investors accept that risk at the same price as institutions. The letter from Warren, the blank figures and the scale of the compute commitments give them plenty to weigh before then.

Source:TNW | Anthropic News