The Anthropic IPO, once pencilled in for October, now looks set for November at the earliest. The lab wants strong third-quarter results in hand before it lists, while still aiming for a valuation near $2 trillion and a raise of up to $100 billion.
Key Takeaways
- The listing is now expected in late October at the earliest, and more likely in November.
- Market participants still expect a valuation around $2 trillion and up to $100 billion raised, both above SpaceX’s June records.
- Rising interest rates, competition from GPT-6 Astra and uninsured cyber risks all weigh on the timing.
Have an AI Sum Up This Article
ChatGPTWhy the Anthropic IPO waits for third-quarter numbers
The timeline moved late last week. Reporting published on Friday, September 18, put the Anthropic IPO in late October at the earliest, with November now the more likely window, a clear slip from the October target the market had been working with.
Advisers framed the delay around a single goal. Anthropic wants to walk into the roadshow with strong third-quarter results, proof that its competitive position holds, rather than list on spring numbers alone.
That reasoning is a little odd, because the second quarter was already a blowout. Revenue went from $4.7 billion in the first quarter to $11.5 billion in the second. Another quarter adds less new information than it adds reassurance that the curve is not flattening.
The paperwork has been moving for months, since Claude’s maker filed with the SEC in early June. By late August the ambition was set too, with the Anthropic IPO targeting a $2 trillion valuation for a listing then expected in September or October.
Those headline numbers have not changed. A $2 trillion valuation and a raise of up to $100 billion would both break the records SpaceX set with its June listing, which came at a $1.77 trillion valuation.
There is also an awkward backdrop. Dario Amodei has been arguing publicly for a more cautious pace of AI development, a stance other tech leaders have echoed, and Horizon covered the moment Claude’s founder said he wants to slow AI down. Investors now have to price a company whose CEO is calling for restraint.
Revenue doubles while the compute bill climbs 65%
The projections behind the Anthropic IPO are enormous. Backers expect $110 billion in annualized revenue by year-end, and figures shown to investors put 2028 revenue at $190 billion to $200 billion.
Set against the spring, the climb is steep. In May the story was Anthropic closing in on a $900 billion valuation. Four months later, the target is more than twice that.
Costs are rising as well, just more slowly. Infrastructure spending went from $3.4 billion to $5.6 billion over the same stretch, up 65%. Revenue outpacing compute costs is what improves the operating margin, and it is the one trend the third quarter has to confirm.
One contract shows the scale. Anthropic’s computing deal with SpaceX costs $1.25 billion a month, the kind of commitment a $100 billion raise is meant to fund. Summer had already set a high bar, when Anthropic’s revenue hit a $65 billion annual pace.
Rates make the math harder. Higher interest rates push up the cost of data centers, and a lab whose growth depends on stacking compute contracts feels every move.
For Claude users, the calendar is not abstract. Public capital pays for the compute behind usage limits, peak-hour availability and the pace of new models. A one-month slip changes nothing today, but a weak listing would hit all three.
Enterprise buyers are watching too. A listed lab reports its accounts every quarter, which gives procurement teams a view of their Claude supplier’s finances they simply do not have today, and makes multi-year commitments easier to sign off internally.
More articles on Horizon
- Claude Now Leads 26% of Anthropic’s AI Research
- AI Actress Tilly Norwood Switches to Chinese Mid-Interview
- AI Hallucination Nearly Sent US Troops Onto a Chinese Ship
OpenAI waits until 2027, and the Anthropic IPO sets the benchmark
Anthropic is not the only lab pushing back. OpenAI has moved its own listing to 2027, citing the safety questions facing companies chasing general AI, while it discusses a private raise at a $1.2 trillion valuation.
A November Anthropic IPO would still land months ahead of OpenAI, giving Claude’s maker the chance to set the public price reference for frontier labs before its rival does. For every other lab raising money, that daily market price would become the anchor of its next negotiation.
Competition is part of the story, though. OpenAI’s September release of GPT-6 Astra is among the factors cited for the delay, and a solid third quarter would show that Claude kept its commercial ground through that launch.
Then there is cyber risk. Safety testing exposed unintended hacks during some exercises, and those operational incidents are not covered by insurance. The final prospectus will have to say plainly how that risk is managed, a disclosure the rest of the sector will study closely.
The broader question hangs over every AI valuation: does heavy AI spending produce measurable returns? A listing at the top of the range would answer yes. A discount would restart the bubble debate across the industry.
Follow the story on Horizon.
