Anthropic publicly filed its registration statement for an initial public offering (IPO), Form S-1, on October 26 U.S. time. The company plans to list on the Nasdaq Global Select Market under the ticker symbol “ANT,” with Goldman Sachs and Morgan Stanley serving as lead underwriters. The public filing follows a confidential filing process that began in June, allowing investors to review the company’s financial performance, operations, and risk factors. According to major media reports, Anthropic is targeting a November listing and could raise as much as $100 billion. However, the number of shares to be offered, the price range, and the final listing date have not yet been determined. Filing a registration statement does not mean the IPO has been completed.
Anthropic Is a Leading Developer of Frontier AI Models
Founded in 2021 by former OpenAI employees, including Dario Amodei and Daniela Amodei, Anthropic is one of the world’s leading frontier AI model developers. Its core products include the Claude model family and applications such as Claude Code and Cowork, which focus on software development, enterprise knowledge work, and AI agent services.
The company’s revenue primarily comes from model services billed based on usage. Customers use Claude through an API or cloud platforms and pay according to the model and number of tokens used. In the first half of 2026, these services accounted for about 80% of total revenue, while paid subscriptions such as Claude Pro and Max contributed the remaining 20%. As of the end of September 2026, more than half of Fortune 500 companies had contributed at least $100,000 in revenue to Anthropic over the prior year, indicating that the company has built a substantial presence in the enterprise market.
In May, Anthropic completed a $65 billion financing round at a post-money valuation of $965 billion. Recent market estimates put its potential IPO valuation as high as $1.5 trillion to $2 trillion. Compared with the current market capitalizations of publicly listed companies, a $1.5 trillion valuation would rank around 12th globally, near Tesla. At $2 trillion, Anthropic could rank sixth or seventh, on par with SpaceX and ahead of Meta and Broadcom. In IPO terms, a $2 trillion valuation would also exceed SpaceX’s roughly $1.8 trillion valuation when it listed this year, potentially challenging the record for the highest IPO valuation in history. These comparisons are based on current market capitalizations; the actual offering valuation, post-listing market capitalization, and global ranking will depend on the final pricing and market performance.
Anthropic’s IPO Filing Shows Rapid Revenue Growth, While Compute Spending and Sustainable Profitability Remain in Focus
Key details from Anthropic’s IPO filing and listing plans include:
Anthropic generated $4.585 billion in revenue in 2025 and $16.286 billion in the first half of 2026, an increase of about 1,300% year over year. As of the end of September, revenue over the trailing 12 months was up about 1,200% year over year. Its operating loss narrowed to $693 million in the first half, and it recorded $974 million in operating profit in the second quarter. Net loss for the same period reached $80.748 billion, of which $72.388 billion came from changes in the fair value of financial instruments and should not be treated directly as cash burn. Operating cash flow was positive at $3.929 billion, though it was also affected by the timing of collections and payments and by changes in working capital.
On financing, Anthropic had raised about $141.1 billion in total as of the end of June, but this does not equal cash available for use. A further $15 billion in conditional financing from Amazon remains available. In March, Google purchased 38.6 million non-voting convertible preferred shares for about $10 billion. Anthropic may also issue up to $42 billion in convertible notes to Broadcom or its designated financial partners, subject to certain conditions, to fund equipment leases; none had been issued as of the filing. IPO proceeds will be used for operations, compute, infrastructure, and talent, and may also go toward acquisitions or investments. The final amount raised and the level of dilution remain uncertain.
As of the end of September, Anthropic had more than 11,900 customers that contributed at least $100,000 in revenue over the previous year. In the third quarter, 79% of revenue came from customers outside its top five, and no single customer accounted for more than 10% of revenue in the first three quarters. Net revenue retention was about 900%, meaning revenue from the same cohort of qualifying customers had grown to nine times its level a year earlier. However, the calculation excludes Pro and Max subscribers. About 48% of revenue in the first half came through cloud platforms. These platforms serve simultaneously as suppliers, distribution channels, and competitors, creating dependency and potential conflicts of interest.
On compute, Anthropic has committed to spend at least $110 billion with Amazon by April 2036 and about $114.5 billion with Google by July 2033. If it does not use enough capacity, it may have to make up the shortfall. The agreement with SpaceXAI, which extends through 2029, has a spending cap of about $120.5 billion. However, after the initial non-cancellable period, the agreement can be terminated with advance notice, so the full amount should not be treated as an irrevocable commitment. In addition, estimated payments of $125 billion to $150 billion for additional TPU capacity to be delivered between 2028 and 2031 are not included in the contractual commitments table and do not represent the company’s total infrastructure spending.
On governance, the seven co-founders will control 50.1% of the voting power on matters subject to a general shareholder vote. The Long-Term Benefit Trust can generally appoint and remove a majority of directors, limiting the influence of ordinary shareholders. Anthropic also warns that its rapid growth may not be sustainable, that expansion will increase costs, and that losses may continue. The company may never achieve or sustain profitability. Investors should watch whether revenue growth can keep pace with compute commitments, as well as risks related to AI safety, cybersecurity, copyright litigation, regulation, and supplier concentration.
IPO Could Expand Funding Flexibility, While Public Markets Test AI’s Commercial Returns
The IPO’s main value to Anthropic would be access to more equity capital to fund long-term research and compute investment. This could help bridge the gap between multi-year minimum purchase obligations and revenue that varies with usage, giving the company more capacity to manage research cycles, demand fluctuations, and early investment in capacity. A listing could also expand its options for future capital raises, support acquisitions and employee compensation through stock, and reduce reliance on financing from specific strategic investors by broadening its investor base. However, dependence on compute suppliers and distribution channels would still need to be addressed through supplier diversification and improved commercial terms. New funding must also translate into higher compute utilization and commercial revenue to generate lasting returns.
For the broader AI industry, Anthropic’s listing would provide a public-market valuation benchmark for frontier model businesses, allowing investors to assess how model capabilities and enterprise adoption translate into shareholder returns. Key questions include whether revenue growth comes from new customers, increased usage by existing customers, or changes in pricing and product mix—and how much revenue remains to fund research after inference compute and distribution costs are deducted. If Anthropic can maintain pricing and use technical improvements to lower costs, growth is more likely to translate into sustained profitability. If efficiency gains are mainly passed on through lower prices, increased usage may generate only limited cash returns. The company’s public-market valuation will therefore depend on whether its technological lead creates sustained customer willingness to pay and how much of the economic value it can retain.
For suppliers, the IPO would first increase the funding certainty of existing procurement and expansion plans; only later might it lead to additional demand. Anthropic has arranged compute resources across different chip and cloud architectures through partners such as Amazon, Google, Broadcom, and SpaceX. A successful fundraising would help these plans proceed as scheduled. However, demand forecasts should be revised upward only when supported by new orders, expanded procurement, or earlier deliveries, to avoid double-counting existing multi-year contracts. The actual benefits to chip, memory, server, and power-equipment suppliers will depend on compute architecture, procurement allocations, delivery terms, power availability, and capacity utilization. They cannot be inferred from the amount raised alone.
Compute Payment Commitments and Revenue Volatility Put Liquidity and Valuation to the Test
Anthropic’s main risk is that revenue fluctuates with usage and pricing, while some compute expenses are subject to multi-year minimum purchase obligations. When demand is strong, higher utilization can help spread costs. When demand falls short, idle compute capacity can erode profitability. Price competition could add further pressure. For example, all else being equal, if average selling prices fall by 20%, usage would need to rise by 25% to maintain revenue, while the additional inference costs could still reduce profitability. Investors should therefore track usage, realized pricing, and unit costs together.
Anthropic’s compute partnerships exceed $600 billion in scale, but they cover different periods and conditions and should not be treated as short-term debt. Investors should focus on payment obligations over the next one to two years and the flexibility to adjust procurement. Revenue is somewhat diversified: the top five customers accounted for about 25% of revenue in the first three quarters of 2026, and no single customer contributed more than 10%. However, many large customers lack long-term spending commitments or have contracts nearing expiration, and can still reduce spending. A net revenue retention rate of about 900% also does not guarantee future demand. If growth slows while payments peak, liquidity could come under pressure. A breach of certain commercial agreements could also trigger cross-defaults, accelerating related debt and further affecting financing and access to compute. Although an IPO could add a financial buffer, investors will still need to monitor established customers’ usage, spending, and implementation benefits to determine whether revenue can support the company’s commitments.
On valuation, even continued growth may fall short of the expectations implied by the offering price. If growth slows, prices fall, or cash collections are delayed, the market could lower both earnings forecasts and valuation multiples. If Anthropic needs to raise more capital, a falling share price would increase the amount of equity dilution required to raise the same funds. Financial analysis should also distinguish accounting earnings from cash generation. In particular, the improvement in operating cash flow in the first half included the effects of collection and payment timing and working capital changes. Future financial statements should be assessed for whether operating improvements can continue and whether cash remains after necessary investment.
Conclusion
Overall, Anthropic’s IPO could broaden its funding sources, support model research and compute investment, and provide a public-market valuation benchmark for frontier AI businesses. The company’s rapid revenue growth, expanding enterprise customer base, and improving core operations show progress in commercialization. Whether these developments can support the high valuation expected by the market will depend on sustained customer spending, pricing power, and the ability to turn compute efficiency gains into stable cash returns. Revenue varies with usage, while some compute payments have already been committed. IPO proceeds could ease financial pressure, but long-term performance will still depend on operating results. Investors should focus on the final offering terms, customer spending trends, and compute payment arrangements, while assessing supplier benefits based on actual new orders and delivery progress.
