Anthropic is about to be "junk"-rated with $112B in cash. I did the Project North Star math.

Anthropic has not been rated yet. That is the first thing to get straight, because the headline makes it sound like a verdict has landed. It hasn't. What happened today is that the Financial Times reported Blackstone — the anchor investor in Anthropic's $63 billion chip financing, Project North Star — has told investors in that deal it now expects Anthropic to receive a "high" junk rating, in the double-B range, after its IPO. TechCrunch picked it up this afternoon, and Seeking Alpha, Invezz, and AI Weekly all corroborate the same FT report with the same numbers. Anthropic and Blackstone both declined to comment. Fitch, Moody's, and S&P have not issued formal ratings.

So this is a prediction from the lender with the most skin in the game, not a rating action. It is also a reversal: Blackstone previously told investors Anthropic could reach investment grade before 2027. Now it expects the opposite side of the line. That reversal is the story — and the deal math explains why it is not a contradiction.

The $63B, split in two

Project North Star is not one bet on Anthropic. It is two very different bets stapled together:

Project North Star total:          $63.0B
Senior loans (Broadcom-guaranteed): >$42.0B  (66.7% of the deal)
Subordinated / junior debt:         $18.0B  (28.6% of the deal)
Blackstone funds' share of junior:  ~$9.0B  (50.0% of the junior slice)
Junior yield (Blackstone slice):    above 10%
Anthropic cash and equivalents:     >$112B
Cash / junior slice:                6.22x
Cash / whole $63B deal:             1.78x

I ran those ratios myself from the reported figures. Read them and the "junk" label starts to look less alarming and more precise. Two-thirds of the deal never depended on Anthropic's credit at all — it rides on Broadcom's investment-grade guarantee. The slice that actually prices Anthropic's standalone credit is the $18 billion junior tranche, and Blackstone owns about half of it, at yields above 10%. That is already junk pricing, whatever the agencies eventually print. The rating would just make it official.

And the cash line is the part the headline buries. Anthropic holds more than $112 billion in cash and equivalents, and investors expect the IPO to raise roughly another $100 billion. Cash alone covers the entire junior slice 6.2 times over. This is not a distress story. Nobody is worried Anthropic misses a payment next quarter.

What the label actually costs

If it is not distress, why does anyone care? Because a sub-investment-grade rating shrinks the buyer pool. Many insurers and pension funds cap how much junk-rated debt they can hold, which means Anthropic's next financing — not this one — gets more expensive and narrower. The arithmetic of that:

1pp extra yield on $18B junior:      $0.180B/yr
1pp extra yield on $100B future debt: $1.00B/yr

Those are illustrations, not forecasts — I am applying one percentage point to reported figures to size the penalty. But they frame the stakes: Anthropic's prospectus describes a compute plan north of $800 billion. If even a fraction of that is debt-funded, each point of rating penalty compounds into billions a year. The rating is not about whether Anthropic survives; it is about what its $800 billion plan costs to finance.

There is also a precedent worth noting, reported rather than confirmed: SpaceX reportedly won investment-grade ratings from all three major agencies days after its June IPO. Anthropic, by contrast, already carries a BB+ from KBRA since June — one notch below investment grade — and a BB from Morningstar DBRS on the junior debt, per the same reporting. The agencies are not starting from neutral here; Blackstone's revised expectation aligns with where the smaller agencies already sit.

The S-1 context, honestly stated

The rating question also puts Anthropic's S-1 in a different light. First-half 2026 revenue was $16.3 billion against a net loss of $80.7 billion for the same period, as reported from the filing. That loss figure is dominated by stock-based compensation mechanics, not cash burn — which is exactly why the cash pile matters more than the loss line for a lender. But a lender still has to underwrite a company whose spending plan is roughly 50 times its half-year revenue run rate. A double-B rating is, in that light, less a judgment on Anthropic's business than a statement about its ambition: the plan is bigger than any single balance sheet, even one with $112 billion on it.

What to watch

Two things, both dated. First, the actual agency decisions after the Nasdaq debut, expected by late November — that is when a prediction becomes a rating. Second, whether the IPO raise lands near the expected ~$100 billion: every dollar of equity raised is a dollar that does not need a rating. If the listing goes well, Anthropic's leverage ratios improve on day one, and Blackstone's double-B call could look conservative within a quarter. If it doesn't, the junk label will have been the early warning, not the noise.

The headline says "junk." The math says: a company with $112 billion in cash, a $63 billion deal that is two-thirds guaranteed by someone else, and a spending plan so large that its financing cost matters more than its current losses. That is not a company in trouble. It is a company whose ambition has outrun the rating scale — and lenders are saying so out loud.

Sources: Financial Times report (October 30, 2026) via TechCrunch (October 30, 2026, 2:52 PM PDT); Seeking Alpha via TradingView, Invezz, AI Weekly, and BigGo Finance corroborating the same FT report with consistent figures; KBRA BB+ rating (June 2026) via Dealroom/BeInCrypto coverage; S-1 figures (H1 2026 revenue $16.3B, net loss $80.7B) as reported in Anthropic's October 26 filing coverage. All ratios computed from the cited figures on October 30, 2026. The double-B expectation is Blackstone's reported guidance to investors, not an agency rating action; Anthropic and Blackstone declined to comment.