Anthropic is running into a familiar problem for fast-growing AI companies: it needs vast amounts of cloud capacity, but its biggest infrastructure suppliers are increasingly also its rivals. A leaked S-1 filing, reported by The Wall Street Journal, says the Claude maker has struggled to secure enough compute from AWS, Google and Microsoft, even as it plans to spend staggering sums on cloud contracts and builds out backup options of its own.
Anthropic’s cloud partnerships are becoming harder to manage
The company’s dependence on hyperscalers has turned into a strategic squeeze. According to the leaked filing, AWS, Google and Microsoft now see Anthropic less as a customer and more as a competitor, creating tension around a relationship that once looked straightforward: Anthropic would buy cloud services, and the providers would supply the infrastructure needed to train and run its models.
That dynamic matters because the company’s needs are enormous. Reuters and the Financial Times had already reported in late September that Anthropic believed it could spend more than $750 billion on cloud contracts, including some agreements that were still being negotiated. The leaked filing reportedly adds more context to those figures, underscoring just how much compute the company expects to consume as it scales Claude and related products.
Even that level of spending may not be enough to guarantee access. The filing says Anthropic has also begun leasing its own data centers and working with third-party data center operators to broaden its supply. But the company is reportedly worried that those operators have a “limited operating history,” making them potentially less dependable than the established hyperscalers that have long dominated enterprise cloud infrastructure.
The scale of Anthropic’s spending is tied to equally large losses
Anthropic’s infrastructure appetite is easier to understand when set against its recent financials. The filing reportedly says the company posted an $80.7 billion net loss in the first half of the year, while generating $16.3 billion in revenue over the same period. The Wall Street Journal also reported that only two customers accounted for about a fifth of sales, a reminder that even a company with rapid growth can still face concentration risk.
Those numbers are eye-catching on their own, but they also reveal something broader about the current AI market: training and serving advanced models is expensive, and the companies at the front of the race are spending at an extraordinary pace to secure the compute they need. For Anthropic, the challenge is not only finding enough capacity today, but also ensuring it can keep expanding without becoming too exposed to any single supplier or customer.
- Reported first-half net loss: $80.7 billion
- Reported first-half revenue: $16.3 billion
- Reported cloud-contract spending potential: more than $750 billion
- Revenue cited for third quarter so far: $16.8 billion
- Quarter-over-quarter revenue growth: 45%
Why the hyperscalers’ role is now more complicated
For years, the major cloud providers were the backbone of AI expansion. They offered the scale, reliability and global footprint required for model training and deployment, while startups and established software firms alike treated them as essential partners. Anthropic’s situation shows how quickly that relationship can become complicated once a customer starts to look like a long-term strategic threat.
Anthropic is not the only AI company navigating this tension, but it is one of the clearest examples of a business whose dependence on cloud providers can collide with those providers’ own AI ambitions. If a hyperscaler is also building its own models, or backing competitors in the same market, the incentive to offer the most favorable capacity or commercial terms may weaken. The result is a more fragile supply chain for the very companies driving the AI boom.
That is especially important for a company like Anthropic, which is reportedly raising capital and preparing for a possible public debut. TechCrunch reported that Anthropic is targeting a November IPO. For prospective investors, the quality and resilience of its compute supply chain is not a side issue; it is central to whether the company can sustain growth, manage costs and meet demand.
Risk disclosures are expanding alongside revenue
The leaked filing does not only focus on infrastructure. It also reportedly builds on earlier disclosures about catastrophic risk, warning that Anthropic could be held liable for damages caused by harmful AI behavior in amounts that could exceed available insurance coverage. That kind of language reflects the growing legal and financial uncertainty surrounding frontier AI systems, especially as they are deployed more widely into business and consumer contexts.
Such warnings are significant because they show how Anthropic’s public story is becoming more complex even as revenues rise. The company can point to strong growth, including the reported third-quarter revenue figure of $16.8 billion and the 45% increase from the previous quarter. But those gains are being accompanied by extremely high spending, operational dependence on a few large infrastructure partners and potential liabilities that could be difficult to insure against at scale.
In other words, the filing paints a picture of a company that is growing quickly but also accumulating a wide range of risks at once. That is not unusual in the AI sector, where demand is strong and competition is intense, but the scale in Anthropic’s case is unusually large and increasingly public.
What Anthropic’s predicament says about the AI boom
Anthropic’s cloud headaches point to a broader structural issue in AI: the companies building the next generation of models are often dependent on the same firms that want to dominate the infrastructure layer. That can create a circular economy in which startups need hyperscalers for compute, hyperscalers want the upside of AI growth, and both sides end up competing for control of the stack.
Leasing data centers and adding third-party operators may help Anthropic reduce dependence on the biggest cloud providers, but those alternatives introduce their own trade-offs. Reliability, maturity and scale all matter when the workloads are as demanding as frontier AI training and inference. The company’s concern about the limited operating history of those providers suggests it understands that compute diversification is necessary, but not simple.
At the same time, the leaked filing suggests Anthropic remains committed to expanding aggressively despite the cost. That means its cloud strategy will likely stay under scrutiny, especially if the company moves ahead with an IPO. Public investors are likely to focus not just on growth, but on how much capital the business must burn to sustain that growth, and whether its infrastructure strategy can keep pace with its ambitions.
For now, the headline is clear: Anthropic may be one of the most prominent names in AI, but the cloud platforms it depends on are no longer uncomplicated allies. As the company scales, the relationship between buyer, supplier and competitor is getting messier by the day.
Source: Original report
Was this helpful?
Explore more: More Business & Startups Tech News Blog
Last Modified: October 21, 2026 at 10:31 pm
0 views
