Blackstone tells investors Anthropic may get a high junk rating after IPO, complicating its $63B chip financing
- Blackstone told investors in Anthropic's $63bn chip financing deal, known as Project North Star, that the company is expected to receive a high junk rating from major credit rating agencies after going public, a reversal from its earlier guidance.
- An investment-grade rating is central to Anthropic's plan because it allows the company to borrow at lower rates; the prospectus filed Monday describes spending more than $800bn on computing infrastructure between now and 2045.
- The North Star financing covers Anthropic's lease of 1.5 gigawatts of Google semiconductors; Blackstone is an anchor investor and is working with the banks that provided the financing to sell the debt on to other investors.
- Anthropic filed to go public on Monday, setting up what could be the largest listing of all time next month.
Hacker News 의견들
The headline is the funny part: it's only a 'may not'. Everyone on the fixed income side already knows these labs price like speculative paper. If there's that much of an equity cushion, the bonds should be investment grade. The fact that they aren't means the bond market doesn't believe the equity valuation at all.
How unusual is this really? Amazon spent more than a decade carrying a junk rating and it worked out fine for anyone who held. Risky is not the same as bad. Every AI lab is a high risk, high reward play and the coupon will simply reflect that.
Sure, but it still shrinks the buyer pool. Some institutions just can't hold junk paper in their mandates, so there is less demand for the debt, which means worse terms for Anthropic. It's not fatal, it's just a smaller set of lenders.
Anyone who has looked at these financials in any normal business sense is not surprised. These labs look terrible on paper. Lending to them is speculative by definition, so the rating is just telling the truth. I've talked to plenty of engineers at work who are genuinely shocked when I say both OpenAI and Anthropic crater.
The dot-com comparison gets overused. Most of those 1990s IPOs were tiny companies raising what today would be a Series C round. Some did die, but you also can't cherry-pick only the failures, or you'd have missed the one that grew 2500x.
I put Anthropic in the worst position of the labs. The AI that actually makes money is the one embedded in phones, cars, TVs, smart homes, and that's Google, Apple and OpenAI powering Siri. Anthropic has none of that distribution.
OpenAI is in the same boat, they don't own distribution either. Retail AI isn't a real business; people are subscription'd out and half of them don't even want the feature. The only AI that matters is the one large corporations pay for, and that is Claude.
The problem is they refuse to raise prices. Demand at these prices isn't sustainable, and unless they put people on the ground at every large firm to rebuild operations around their models, they'll end up in a price war. Cheap Chinese replicas cap the whole thing anyway.
Anthropic won't even let people build video games with it. They're on the wrong side of history. Open weight Chinese models have no such limits, there's no moat, and no, they can't police the whole world. We'll route around it.
The article copy says their plan is 'bold but sane'. That phrasing reads like Claude wrote the prospectus.
An IPO is mostly an exit. The underwriters and early institutional investors get paid first, then founders, then early employees. By the time the retail crowd piles in, the smart money is already out the door.
Popcorn is already stashed. I genuinely want to watch how this one plays out.