Anthropic, OpenAI See IPO Targets Shrink By Up To $800B

Anthropic's targeted IPO valuation has fallen by as much as $800 billion since early October while vertical bets like CesiumAstro and Zipline keep raising at full value.

By the Numbers

~$2T
Anthropic IPO target, early Oct
$1.2T-$1.3T
Anthropic IPO target, Nov 6
$1.4T to $1.25T
OpenAI IPO target cut
$730M at $4.5B+
CesiumAstro Series D, Nov 6
$1.5B at $21.5B
Zipline raise, Nov 5
Trace Cohen
By Trace Cohen Β· Value Add Pulse IPO Desk
Founder & Editor-in-chief, Value Add VC
3 min read
Anthropic, OpenAI See IPO Targets Shrink By Up To $800B: key figures
ShareXLinkedInEmail

THE RUNDOWN

1

Anthropic's targeted valuation has fallen by up to $800 billion since early October, while OpenAI's own target dropped from $1.4 trillion to $1.25 trillion.

2

Vertical, revenue-generating bets like CesiumAstro and Zipline are still closing megarounds at full or rising valuations despite the frontier-lab pullback.

3

General Catalyst alone led $5.1 billion across nine October deals, showing capital hasn't left venture -- it has rotated toward defensible, infrastructure-heavy bets.

4

The gap suggests public-market investors are pricing frontier AI labs' infrastructure spending and thin margins differently than they price picks-and-shovels startups.

The VC Read

Value Add VC analysis

The reset isn't a verdict on AI -- it's a verdict on two unprofitable labs asking public investors to underwrite hundred-billion-dollar compute commitments. GPs should watch whether CesiumAstro and Zipline's investors get rewarded for picking defensible, cash-generating niches over the raw model race, because that's the comparison this gap is really drawing.

Analysis

As of November 7, 2026, two of AI's most-watched IPO candidates have cut their targeted valuations -- Anthropic's target alone has fallen by as much as $800 billion since October, with OpenAI's separately down $150 billion -- even as narrower, revenue-generating startups in defense and logistics keep closing megarounds at rising marks. The split is some of the clearest evidence yet that public-market investors aren't treating every AI-adjacent valuation the same way (full methodology at Pulse's stats page).

Anthropic's targeted IPO valuation has fallen from about $2 trillion in early October to $1.2 trillion-$1.3 trillion as of Pulse's November 6 coverage, a cut of roughly $700 billion to $800 billion in five weeks (previously covered). OpenAI's own IPO target has moved in the same direction, down to $1.25 trillion from $1.4 trillion, according to the same Pulse reporting. Both companies are racing toward public markets while still posting steep operating losses and committing hundreds of billions of dollars to long-term AI infrastructure.

Why frontier labs are getting a discount

The pullback traces in part to AI token spending, which grew just 1.5% between August and September -- a slowdown Pulse has reported is weighing on how bankers and institutional investors model future AI-lab revenue. Anthropic's own amended S-1 disclosed the mismatch bankers are now pricing around:

β€œOpenAI's own IPO target has moved in the same direction, down to $1.25 trillion from $1.4 trillion, according to the same Pulse reporting.”

  • Revenue -- nine-month revenue up 1,206% year-over-year to $33.2B.
  • Operating loss -- a nine-month operating loss of $1.3B.
  • Infrastructure commitments -- long-term commitments of $895.4B.

That combination -- triple-digit growth against a widening loss and near-trillion-dollar infrastructure obligations -- is what makes public investors nervous about a trillion-dollar-plus sticker price.

Vertical bets aren't seeing the same discount

Compare that with two vertical, revenue-generating startups that priced at or above target in the same week:

  • CesiumAstro -- $730M Series D above a $4.5B+ valuation (Nov 6), pushing its 2026 capital raised past $1.2B for a 737-satellite defense-communications constellation. Previously covered.
  • Zipline -- $1.5B raise at a $21.5B valuation (Nov 5), nearly tripling its January mark for its drone-delivery network. Previously covered.

Neither company is anywhere near IPO-ready, but both are getting full or rising private-market pricing while two of the best-known AI labs in the world see their IPO targets cut.

General Catalyst alone led or co-led nine funding rounds worth $5.1 billion in October -- more than any other U.S. venture firm by capital deployed -- a sign that money hasn't left venture investing so much as it has rotated toward companies with defensible niches, government contracts, or physical infrastructure moats, rather than general-purpose model labs competing on compute spend alone.

What the discount doesn't prove

None of this means Anthropic or OpenAI are in trouble -- however, the risk that remains is about pricing, not fundamentals: a $1.2 trillion IPO would still rank among the largest in history, and both companies' revenue growth rates dwarf almost anything else in tech. What the valuation cuts actually show is narrower -- public-market investors are pricing in the cost of the infrastructure commitments and thin margins that come with being a frontier lab, while private investors are still willing to pay full price for startups whose revenue and moat are easier to underwrite. For GPs and LPs, the diligence question worth asking isn't whether AI is overvalued broadly -- it's whether a given company's valuation is being priced against its cash-generating business or against the promise of the model race.

Watch whether Anthropic's roadshow, expected to start as soon as next week, produces a final price closer to the $1.2 trillion floor or the $2 trillion ceiling -- that gap will tell GPs more about where public-market sentiment on AI infrastructure spending is heading than any single funding round.

ShareXLinkedInEmail

More from the Pulse archive: IPOs & Public Markets coverage β†’ Β· All topics

Key Sources

1 source

THE WIRE in your inboxβ€” Tech, startup & VC news with The VC Read, a few times a week. Free to subscribe, no spam.

No spam, unsubscribe anytime. Sent via Substack Β· Privacy Β· or subscribe on Substack β†—