Situation report: Anthropic is reportedly close to doubling down on one of its biggest infrastructure bets.
Anthropic and AI infrastructure provider Nscale are said to be on the verge of announcing another $45 billion compute deal, according to TechCrunch AI, which cites a Bloomberg report. The two companies announced their first $45 billion agreement a little over a month ago. If the second one closes, the partnership could be worth $90 billion in total.
Bloomberg’s sources say talks are still underway and the terms could change. TechCrunch AI has contacted both companies for comment.
🎯 Key Facts
- What: A possible second $45 billion compute agreement between Anthropic and Nscale.
- Who: Anthropic, the company behind Claude, and Nscale, a UK-based provider of AI infrastructure.
- When: Talks are active now. No deal has been signed or announced.
- Status: Unconfirmed. People familiar with the matter say the details could still shift.
- Pattern: This would be the latest in what TechCrunch AI calls “a long string of giant compute deals” Anthropic has struck over the past year.
📡 Why This Matters
To me, the speed is the real story. Big compute partners usually take time to show they can deliver before a customer commits again. Going back to the same provider for another $45 billion in about five weeks tells you how badly Anthropic needs capacity. It also suggests the first deal is going well enough to justify a second one.
Compute has become the main constraint in frontier AI. Training new models and serving millions of users both need huge amounts of GPUs, power, and data center space. Whoever locks up capacity early gets to ship faster and scale further. Anthropic has been buying across several suppliers, including the big cloud providers and specialist infrastructure firms, and Nscale is now one of its largest partners.
It matters for Nscale too. Two deals of this size would make Anthropic an anchor customer, and that changes how investors read the company.
⚠️ Threat Assessment: The Money Question
Everyone isn’t cheering. TechCrunch AI notes that the sheer amount of money flowing into AI compute has “spooked a certain segment of investors,” who worry companies are spending too much for too little profit.
It’s a fair worry. These commitments run years ahead of the revenue meant to pay for them. If demand for AI services keeps growing, the spending looks smart. If growth slows, a lot of companies will be stuck with expensive contracts and empty capacity.
Anthropic is betting it’ll pay off. TechCrunch AI describes the company as driven by the belief that “it will all pay off in the end.”
📈 The IPO Angle
Both companies are pursuing public listings, so the timing matters.
- Anthropic is still expected to go public this year. Big compute deals show investors how ambitious its growth plans are. They also show how much cash it will keep burning.
- Nscale is also seeking an IPO. On Monday, Bloomberg reported that the company is getting investor pushback on its proposed $30 billion valuation.
That second point matters. A fresh $45 billion commitment from Anthropic would give Nscale a strong counterargument. It’s hard to call a valuation too high when one of the biggest AI labs keeps signing bigger checks with you. It also raises a question public market investors will ask: how dependent is Nscale on one customer?
🧭 What to Watch
- Official confirmation. Look for whether the deal gets announced and whether the final number stays at $45 billion.
- Deal structure. Watch for details on timelines, GPU types, and data center locations.
- Nscale’s IPO pricing. A confirmed deal could change the valuation conversation.
- Anthropic’s filing. IPO paperwork would show its total compute commitments for the first time.
- Investor mood. Each new mega-deal adds to the debate over whether AI infrastructure spending has gotten ahead of itself.
🔭 Outlook
For anyone building on Claude, more compute usually means better availability, higher rate limits, and room for bigger models. For everyone else, it’s another sign that the AI race now runs on infrastructure as much as research. Expect more deals like this as labs fight for capacity ahead of their IPOs. Full details are in the original TechCrunch AI report.