Anthropic’s IPO Tests Whether AI’s Circular Financing Can Become Sustainable
Table of Contents
You might want to know
- Can Anthropic’s IPO reduce its reliance on supplier-investors, whose investments and sales are closely intertwined?
- Does the company’s growing revenue provide a credible path to paying its enormous compute commitments, even as AI competition intensifies?
Main Topic
Anthropic’s arrival on public markets represents a major test for the financing model that has helped drive the artificial intelligence boom. The company’s IPO is expected to raise billions of dollars, while also creating a potential exit or liquidity opportunity for some of its existing investors. Among the most consequential are Amazon and Google, which are both investors in Anthropic and major providers of the computing infrastructure it uses.
At an updated IPO price of $83 per share, Amazon and Google together hold stakes valued at around $307.9 billion. Amazon invested $18 billion in the AI company, and its 14.7% stake would be worth upwards of $159 billion. Google’s 13.7% holding is valued at nearly $149 billion, according to ownership figures in Anthropic’s IPO prospectus. These valuations illustrate how much the fortunes of large technology companies have become connected to the businesses they help fund.
Other companies are also positioned to benefit. Nvidia, AMD, Microsoft, Broadcom, Micron, and TSMC joined Google in committing $26 billion to Anthropic at its IPO price through a private placement. Google has invested $12.3 billion in the company since 2023 and has promised another $30 billion, contingent on Anthropic reaching certain performance milestones. Apart from TSMC, these supplier-investors have also announced multi-billion-dollar strategic partnerships with Anthropic that involve purchases of their products.
This overlap between investor and supplier relationships is often described as circular financing. The same companies that provide capital to an AI developer can also earn revenue by selling it cloud capacity, chips, networking equipment, memory, or other infrastructure. In that arrangement, investment funds help an AI company expand, and some of the money can then flow back to its investors through purchases. From a distance, the structure may appear to generate exceptional fundraising for the AI company and exceptional revenue for its suppliers, even though some of the capital is moving between connected parties.
That pattern has contributed to skepticism about whether the AI boom reflects durable demand or a self-reinforcing funding cycle. The concern is not simply that large companies invest in AI; it is that their returns may depend partly on the very spending those investments make possible. The central question is whether Anthropic can ultimately fund its infrastructure needs through revenue from a broad base of customers, rather than relying on continued support from the suppliers that also own shares.
The IPO could mark a meaningful shift in that arrangement. Investments made while Anthropic was private may have generated substantial value, and public trading could make some of that value liquid. At the same time, the company may be able to rely more on public-market investors and corporate bonds instead of asking suppliers and private investors to finance its growth. Those sources of capital do not eliminate financial risk, but they may reduce some of the conflicts associated with supplier-investors funding a customer that then spends money on their products.
Anthropic’s financial position at its expected debut is substantial. The company is expected to have about $220 billion in cash, cash equivalents, and short-term investments, including proceeds from the IPO and the $26 billion private placement. It can also use stock to meet future commitments. Joey Brookhart, who leads the Tokenomics team at SemiAnalysis, estimates that Anthropic needs roughly $280 billion in cumulative revenue over the next four years to pay for its compute bill. Brookhart has described that target as “achievable given where they will be next year from a revenue perspective.”
Brookhart argues that Anthropic’s supplier-investors may no longer need to support the company financially in the same way. “The ownership stake is very worthwhile, and those companies no longer need any more support from [suppliers],” he told TechCrunch. “Amazon doesn’t need to prop them up anymore. The revenue is all coming from other customers.” The argument depends on whether Anthropic can keep attracting customers and converting its position in generative AI into recurring commercial revenue.
Even with a strong cash position, Anthropic’s infrastructure costs remain a significant obligation. A regulatory filing says the company has committed to spend $900 billion, primarily on compute and infrastructure. Some agreements include termination provisions, but Brookhart considers Anthropic well-positioned to meet its obligations. The scale of those commitments means that liquidity alone is not enough: the company must also sustain demand and manage the cost of serving and training its models.
Revenue figures offer one indication of how the company might cover those costs. Anthropic generated $17 billion in third-quarter 2026 revenue, putting it on pace for a $68 billion annualized rate for the year. If the company continues along its current trajectory, Brookhart believes it could cover the gap needed to fund its compute. This projection is not a guarantee; it relies on growth continuing in a market where technologies, prices, and customer preferences can change quickly.
Gil Luria, head of technology research at D.A. Davidson, also sees the IPO as giving Anthropic a large financial cushion. He told TechCrunch that the company would have more cash on hand than any company had ever had by the time its public offering was complete. Luria said Anthropic raised around $141 billion in capital and would take home around $91.7 billion in net proceeds at $83 per share. His characterization is that the company would hold a “war chest,” rather than simply burning through cash as it raises funds.
A public listing may also make Anthropic a more credible borrower, potentially improving its ability to raise money outside public markets, Brookhart said. That possibility matters because reports have suggested that creditors may not view Anthropic as “investment-grade” material. Public status can provide greater visibility into a company’s finances, though it does not by itself ensure that lenders will offer favorable terms or that future borrowing will be appropriate.
Luria distinguishes Anthropic from neocloud companies such as CoreWeave, which he regards as more dependent on circular financing. “CoreWeave wouldn’t exist if Nvidia didn’t prop them up, and it probably won’t exist if Nvidia doesn’t bail them out,” he said. Brookhart, meanwhile, rejects the idea that Anthropic’s financing is purely circular, describing it as still “additive to the ecosystem.” Their differing assessments underline an important distinction: the presence of supplier-investors does not, on its own, prove that a company’s entire business model depends on recycled capital. The durability of customer demand and the company’s ability to meet obligations are also central.
For both analysts, the most significant risk is that Anthropic could lose its position in the AI race. Its growth is currently supported by its standing at the frontier of generative AI. That advantage could weaken if OpenAI, Google DeepMind, or lower-cost open-weight models become more attractive to developers and businesses. Luria summarized the downside in stark terms: “If we’re all using other people’s models, then Anthropic’s not worth a trillion and a half dollars. It’s worth nothing.” The quote captures the sensitivity of a high valuation to assumptions about future leadership and continued adoption.
However, the outlook for infrastructure suppliers may be less dependent on which particular model provider wins. If businesses and individuals use more AI overall, demand for data centers and computing could continue to grow even if customers switch between models. Luria said the risk for Amazon and Google is not especially high in that scenario because users might simply choose other models. He expects substantially more AI use, while acknowledging that it may not all involve Anthropic. That activity would still rely on data centers managed by companies such as Amazon, Google, and Microsoft.
Debate over cheaper tokens and increased competition also affects estimates of Anthropic’s future profitability. Lower prices can make AI more accessible and increase usage, but they may also reduce revenue per unit of compute. Whether higher volume offsets lower prices depends on customer growth, the mix of services, infrastructure efficiency, and the cost of training and serving models. As a result, revenue growth should be assessed alongside margins and cash requirements, not treated as a complete measure of financial health.
Anthropic’s IPO therefore does not settle the debate about circular financing. Instead, it changes the context in which the question can be examined. The company may enter public markets with a large pool of capital, strong reported revenue, and a path to financing beyond supplier support. Yet its spending commitments are enormous, and its ability to meet them remains tied to continued commercial adoption and its competitive position. The IPO’s significance will depend not only on the money raised, but also on whether the business can turn investment-fueled expansion into durable, independently supported demand.
Key Insights Table
| Aspect | Description |
|---|---|
| Supplier-investor relationships | Amazon, Google, and other infrastructure firms invest in Anthropic while also supplying products and services it buys. |
| IPO and private placement | At $83 per share, the IPO and $26 billion private placement are expected to contribute to an estimated $220 billion in cash, cash equivalents, and short-term investments. |
| Revenue trajectory | Anthropic generated $17 billion in third-quarter 2026 revenue, equivalent to a $68 billion annualized pace for the year. |
| Compute commitments | The company has committed to spend $900 billion, primarily on compute and infrastructure, and Brookhart estimates it needs roughly $280 billion in cumulative revenue over the next four years to pay its compute bill. |
| Competitive risk | Anthropic’s prospects depend on maintaining its position as a leading AI provider amid competition from OpenAI, Google DeepMind, and open-weight models. |
| Infrastructure outlook | Even if customers choose other AI models, data-center operators and infrastructure suppliers could benefit from broader growth in AI use. |
Afterwards...
Anthropic’s public listing creates an opportunity to examine how AI companies finance growth when their investors are also among their most important suppliers. Future analysis should look beyond headline valuations and capital raised to track customer concentration, recurring revenue, operating costs, and the share of spending directed toward related suppliers. Clearer disclosures about contractual commitments and financing arrangements would help investors distinguish genuine end-user demand from activity supported primarily by connected capital.
Several technology and knowledge areas merit continued attention. More efficient chips, improved model architectures, and better data-center energy management could reduce the cost of delivering AI services. Research into how inference costs change as models become more widely used could clarify whether lower prices expand demand enough to offset lower revenue per unit. Independent evaluation of model performance, reliability, and total operating cost could also help customers make choices based on practical value rather than market momentum alone.
Competition among frontier systems and open-weight models will remain central. Researchers and policymakers should explore how transparency, safety evaluation, data governance, and secure deployment can advance alongside performance. The long-term test is not simply whether AI companies can attract capital, but whether they can create useful systems whose costs, revenues, and infrastructure demands remain sustainable. Anthropic’s IPO offers one important case study, but the broader outcome will depend on the evolution of the entire AI ecosystem.
Last edited at:2026/11/19
