Anthropic IPO risk centers on revenue quality, veteran says
A former Goldman Sachs and JPMorgan executive said Anthropic’s biggest IPO risk is revenue quality, citing a mismatch between contract-based costs and usage-based revenue.
The biggest risk after Anthropic's IPO would be its "revenue quality," which is a structural mismatch in the business that may be hidden by the current run-rate, says Dr.
Chan Ahn, Founder and CEO of Tessera PE and a former Goldman Sachs and JPMorgan executive.
Dr Ahn told in an exclusive interview that the cost side of Anthropic's business runs on contracts, while the revenue side is based on usage.
This mismatch could affect how public markets value the company.
The company's revenue is metered and continuously repriced based on both volume and unit price, with no contractual term.
Anthropic has a fixed commitment of roughly $1.25 billion a month running until May 2029, which amounts to about $15 billion a year or close to 30% of its estimated 2026 revenue.
This commitment is to Space Exploration Technologies Corp. (NASDAQ: SPCX ), a direct competitor, after it acquired xAI.
Read Also: Salesforce Stock Rockets 20% on Anthropic Windfall, Guidance Raise The Wall Street Veteran compared Anthropic's situation to the long-haul bandwidth build-out from 1999 to 2002, where fixed long-dated capacity commitments were sized to extrapolated demand curves, and unit prices deflated as capacity arrived faster than traffic.
The same could happen to Anthropic, turning its capacity commitment into a stranded cost.
Dr.
Ahn stated that the single most important figure from the S-1 filing would be the split between committed-spend enterprise agreements and pure on-demand usage.
This would reveal whether the company is funding a fixed multi-year obligation with a revenue stream that could be interrupted in less than three weeks.
Regulatory Shock Tests Revenue He also pointed out the risk of the company's revenue stream being interrupted by regulatory action.
In June, U.S. export controls temporarily pulled Anthropic's two leading models offline for 19 days, reportedly slowing revenue growth before it recovered.
The episode highlighted the risk of Anthropic's usage-based revenue model, which can quickly respond to disruptions, while its fixed contractual costs remain unchanged. "A contracted enterprise book does not bend inside nineteen days.
A metered service with cancellable demand does.
The obligation, meanwhile, did not move at all," said Ahn.
AI Labs Carry Research Risk He stated that the key structural concern is that frontier AI labs are financing high-risk scientific research with venture capital, even as they are being valued like mature software companies.
Historically, similar research risks were largely absorbed by governments or protected monopolies, rather than public-market investors. "Listing does not remove the science-tranche risk.
It transfers it to a holder who marks to market daily," Ahn said.
SpaceX Offers Anthropic a Warning Anthropic's reported $2 trillion valuation faces a reality check as Ahn warned that the company must deliver extraordinary revenue growth and margin expansion to justify the figure.
Ahn estimates Anthropic would need roughly $725 billion in revenue by 2036, or about $950 billion under a higher discount rate.
While its annualized revenue run rate has surged to $65 billion, profitability remains a concern, with its projected Q2 operating margin at just 5.1%.
The Wall Street veteran pointed to SpaceX's post-IPO experience as a warning, arguing that public markets impose far greater earnings scrutiny than private investors.
He also says Anthropic's valuation is largely a bet on the broader AI sector, with enterprise workflows providing a more durable advantage than model leadership alone Earlier this week, a report suggested that Anthropic is preparing to target a potential revenue opportunity of more than $30 trillion, surpassing SpaceX's estimated $28.5 trillion market opportunity.
The Claude maker could seek to raise as much as $100 billion and is expected to file IPO disclosure documents in the coming weeks, with a public listing possible as early as September or October.
Read Also: OpenAI Rallys 100+ Companies for a Global Cyber Defense Push Photo courtesy: Shutterstock