Analysis
Value Add Pulse tracks $201.7 billion in disclosed venture capital across 3,852 stories, and this week's run of deals shows a widening fork in how capital markets are pricing the AI boom: equity investors are still writing checks at record valuations, even as credit-rating agencies and bond desks turn sharply skeptical of the debt propping up AI's physical buildout.
Equity investors haven't blinked
The private funding market kept pricing AI and adjacent deep-tech companies up through the week, even as public AI stocks fell:
- Castelion -- valued at $13 billion since its August Series C, the hypersonic-missile startup has grown past 400 employees racing to open a new production site (Pulse).
- Cohere -- targeting a CAD $35.7 billion valuation in a new Series E, backed by 165% year-over-year ARR growth and 20x customer growth in a year (Pulse).
- Dulo -- the AI hardware-design startup Sebastian Thrun quietly launched this year, raised a $130 million seed from Eric Schmidt, John Doerr, DST Global and Lowercarbon Capital.
- Crosby -- the AI-native law firm, has raised $85 million at a $400 million valuation while bidding for the wind-down assets of Weil, Gotshal & Manges.
- Foundry Robotics -- raised an oversubscribed $100 million Series A to open a 130,000-square-foot autonomous factory in Emeryville, California.
- Group14 Technologies -- has now raised more than $1 billion in total equity for its silicon battery materials business, after a $169 million Series D extension closed October 30.
“- Crosby -- the AI-native law firm, has raised $85 million at a $400 million valuation while bidding for the wind-down assets of Weil, Gotshal & Manges.”
None of those rounds show investors pulling back on price. If anything, Castelion's and Cohere's numbers argue the opposite: capital is still chasing category leaders at multiples that assume years of continued growth.
Debt markets are telling a different story
The same week, the credit side of AI financing flashed real warning signs:
- Anthropic -- has lined up roughly $138 billion in debt-backed financing since September, on top of $6.2 billion in debt already on its balance sheet as of its S-1 (Pulse).
- Amazon -- disclosed a supplier-financing facility of up to $80 billion to underwrite its own AI buildout.
- Intel -- stock fell 28.6% on October 30, wiping out roughly $12 billion of the value of the U.S. government's stake in the company.
- Anthropic (IPO rating) -- Blackstone now expects a double-B "junk" credit rating at its IPO, a downgrade from its earlier investment-grade forecast (Pulse).
OpenAI's own fundraising shows the squeeze directly: the company has cut its valuation target from $1.4 trillion to $1.25 trillion this fall (Pulse).
Its fundraising target has fallen just as fast, from an original $30 billion to just $20 billion this week, with only about $11 billion secured so far (the target was cut further the same day).
The caveat behind the gap
The disconnect is real, but it deserves a caveat: private valuations like Castelion's $13 billion and Cohere's CAD $35.7 billion target are marks set by existing investors and insiders, not prices discovered in a liquid market. They are untested by the same forces now repricing AI-linked debt and public AI stocks, and private marks have historically lagged public multiple compression by two to four quarters rather than moving in lockstep.
It's also worth noting that not every part of the equity market shows AI-sized caution: three IPOs priced between October 27 and 30 -- OPay, Bambusa Therapeutics and Aspen Neuroscience -- raised a combined $613.5 million without any AI premium attached, suggesting equity markets for non-AI assets are functioning normally even as AI-specific credit gets questioned.
The next data point
Anthropic's actual IPO pricing, and the rating it receives from S&P, Moody's or Fitch, will be the first hard test of whether private AI marks hold or start to reprice down toward what credit markets are already signaling. A confirmed junk rating, paired with continued record private rounds, would mean the two markets are pricing genuinely different risks rather than converging on one view of AI's financing health.