IPO Watch
Anthropic's Mega-IPO Meets
Investor Skepticism
Reports say investors are pushing back on Anthropic's planned mega-IPO with pointed questions this August. A $965 billion valuation in May and a confidential S-1 filing in June looked like a straight line toward a record listing — now that "scale" story is being pressed into a "pricing" one.
From a "scale" story
to a "pricing" story
The story starts on May 28, 2026. Anthropic closed a $65 billion Series H-1 round that put its post-money valuation at $965 billion — nearly $1 trillion — surpassing OpenAI's most recent reported valuation and prompting headlines about the "world's most valuable private AI company." Days later, on June 1, Anthropic itself confirmed it had confidentially submitted a draft S-1 to the SEC, setting the IPO machinery in motion.
By mid-July, multiple outlets reported that preparations for an institutional roadshow were underway. Everything up to that point reinforced the "record-breaking IPO" narrative — but the tone shifted in August. The Decoder and other outlets reported that preliminary investor meetings were surfacing pointed, skeptical questions. If May and June were about scale, August looks like the moment investors started demanding concrete justification for a $965 billion price tag.
| The May–June mood | What investors ask in August |
|---|---|
| Round size and valuation take center stage | Does performance justify a $965B valuation? |
| Confidential S-1 filing as procedural progress | Competitiveness against cheap Chinese models |
| A "record IPO" narrative built on scale | Tension with Washington, data-center backlash |
"If I was the head of Anthropic or OpenAI, I'd be petrified."
—Steve Eisman, the investor known as the model for "The Big Short" (via Benzinga)
Sizing up the bet
The gap between valuation and revenue, and the pace of the Chinese catch-up — the two things investors keep asking about.
Anthropic's most recently disclosed annualized revenue run-rate was $47 billion, as of the May 28 Series H announcement. Third-party trackers cited by CNBC and others put late-July run-rate closer to $69–74 billion. That is fast growth by any measure, but a $965 billion valuation still implies a multiple in the low double digits on trailing revenue for a company that remains unprofitable. An estimated $19 billion a year in compute spend is another line item investors keep asking Anthropic to explain on the path to breakeven.
Three arguments
investors keep raising
These are the three points reportedly named directly in roadshow meetings.
Cheap Chinese models
Chinese labs' models — MoonshotAI's "Kimi" line and Alibaba's "Qwen" line among them — have rapidly closed the performance gap on major benchmarks. As CNBC has reported, the six-to-twelve-month lead frontier labs once claimed can no longer be taken for granted, and investors know it.
Tension with Washington
In April, the Pentagon labeled Anthropic a "supply-chain risk" after a dispute over contract terms, and export controls were later applied to some Claude models. As TechCrunch and others have reported, Anthropic has stuck to an independent line while most large tech companies have aligned with the administration — a stance some see adding uncertainty to the IPO review process.
Data-center backlash
Local opposition to new AI data centers — over electricity costs and environmental impact — has spread in multiple regions. Whether Anthropic's and its peers' infrastructure buildout plans proceed as planned is one of the questions repeatedly raised in investor meetings.
Who it affects, and how
This tug-of-war matters most directly to decision-makers at companies running Claude in production. How Anthropic's capital base and ownership structure evolve has real bearing on pricing stability and long-term support. If the IPO slips or shrinks, it's worth confirming now that your stack can lean on a multi-vendor setup rather than a single model provider.
Product managers building on the Claude API should watch just as closely. A pre-IPO Anthropic under pressure to show a path to profitability could tighten free tiers or revise pricing. On the other hand, intensifying competition from cheap Chinese models could push pricing the other way. Either direction is plausible right now, so it's worth building both into your roadmap assumptions rather than assuming stability.
Three milestones to track
These three will largely determine how the IPO actually plays out.
The public S-1
When the confidential draft converts to a public filing, previously undisclosed financials — losses, ownership structure — become visible for the first time.
Roadshow reception
Heading into the reported September–October listing window, how much institutional demand shows up in meetings will drive the final pricing range.
Chinese benchmark trends
How much further Moonshot's, Alibaba's, and other Chinese models close the gap keeps directly testing Anthropic's "frontier edge" pitch.
Skepticism isn't necessarily fatal
Still, tough questions in a roadshow are not, by themselves, unusual — institutional investors interrogating a valuation is closer to standard practice for a mega-IPO than a red flag. Anthropic's annualized revenue is estimated to have grown from $47 billion in May to over $69 billion by late July, and gross margin is estimated to have improved from roughly -94% around 2024 to the 40–60% range in 2026. The underlying growth trajectory looks less shaky than the skeptical headlines might suggest.
Anthropic isn't sitting still, either. It has reportedly made AI in healthcare and biology a centerpiece of its investor pitch, aiming to rebuild public trust and diversify revenue at the same time. On Chinese competition, the company is said to be telling investors it will stay focused on frontier-tier models rather than chase price. The wave of skeptical coverage is real, but it's better read as the market starting to demand accountability for the valuation — not proof the IPO is doomed.