VideoPatrick BoylePatrick Boyle

Anthropic Just Filed to Go Public. Does the Math Add Up?

The presenter questions whether Anthropic’s reported roughly $2 trillion valuation is supported by its revenue, emphasizing that much of the price depends on future growth and that private-market pricing has limits. The discussion also examines data-center financing and construction risks, weak IPO conditions, and the gap between AI market estimates and realized outcomes. Falling model costs and rapid usage growth sit alongside questions about production-level gains, while Nvidia’s valuation and margin outlook offer a public-market comparison.

Artificial IntelligenceMarket ValuationsVenture CapitalTechnology StocksRisk ManagementEquities

Preview · 14 more points in the app

Read the Full Summary

Key Takeaways

The presenter argues that Anthropic’s roughly $2 trillion valuation—about 31 times revenue—depends overwhelmingly on future growth: even an unusually favorable calculation that assumes no costs, staff, or taxes puts present value at about $1.27 trillion. Morgan Stanley’s estimate of a $60 trillion generative-AI market and Anthropic’s internal scenario of more than $10 trillion in added U.S. GDP by 2030 are potential-market estimates, while the presenter’s Uber and WeWork examples caution that a large addressable market does not ensure a company captures it.

The presenter says Anthropic’s private-round valuation is shaped by cloud giants that benefit when it rises, while investors unable to bet against a stock may give optimistic investors disproportionate influence. An IPO would test the public-market value of a pure AI lab and put a market price on Amazon’s and Google’s paper gains; those stakes accounted for 85% and 87% of the companies’ respective net income in their latest results.

Higher rates pressure AI infrastructure from both sides: Matt Kennedy says they lower the present value of distant profits and raise data-center borrowing costs, while the presenter links the 10-year Treasury yield reaching 5.23% to sticky inflation, a hot economy, and a large deficit. The presenter says SB Energy’s hoped-for $50 billion valuation equals roughly 400 times annual EBITDA, despite its prospectus showing $174 billion needed for promised construction; facilities and GPUs are not depreciated until they are ready and switched on.

SB Energy faces project and financing risks as it builds capacity for AI customers: 71% of Americans surveyed opposed a data center near them, and its filings warn that a delayed project may in some cases be bought outright by a customer at a low price. Meanwhile, SoftBank is marketing more than $11 billion in bonds at a 9.75% yield to fund another OpenAI payment, OpenAI expects to burn almost $280 billion by the end of 2030, and its planned 20-year lease of SB Energy’s Ohio campus would provide revenue to the data-center company.

The IPO market described in the episode was subdued: only three IPOs had occurred since Labor Day, five of the year’s 10 largest listings were below their offer prices, and WholeTech postponed its IPO citing market conditions. Against that backdrop, the presenter says Anthropic’s IPO would be the first public-market test of what a pure AI lab is worth; SB Energy had filed before Labor Day but had not begun marketing shares.

AI-ECONOMICS

TAM Versus Outcomes — The presenter contrasts Uber’s $122.3 trillion total addressable market claim at its 2019 listing with annual revenue below $60 billion. He also notes that WeWork went bankrupt despite claiming a $3 trillion addressable market. These examples show that a large stated market does not ensure a company will capture it.

“Uber touted a 122.3 trillion TAM when it listed in 2019. Its annual revenue today is under $60 billion. So they still have 99.5% of their market ahead of them…”—

Musk’s AI Prediction — Elon Musk has predicted that AI and robots will make work optional and money irrelevant.

“Elon Musk has predicted that AI and robots will make work optional and that money will stop being relevant.”—

AlphaEvolve Advance — Google DeepMind’s AlphaEvolve found a better way to multiply certain matrices after the previous method had been used for 56 years. The presenter notes that the impressive results involved tests with a right answer, rather than establishing broad recursive self-improvement.

“Last year, Google Deep Minds Alpha Evolve found a better way of multiplying certain matrices than the one mathematicians have been using for 56 years, which is impressive…”—

ANTHROPIC

Valuation Test — Anthropic is reportedly seeking a roughly $2 trillion valuation, about 31 times its revenue and comparable to the combined value of the 10 biggest tech IPOs. The presenter argues that even with no costs, staff, or taxes—and discounting all revenue at the Treasury rate—its present value would be about $1.27 trillion. He says almost all of the $2 trillion valuation is therefore a bet on growth.

“Anthropic is reportedly seeking a valuation of around $2 trillion, which The Economist points out is about as much as the 10 biggest tech IPOs of all time added together.”—

Valuation Challenges — The presenter says standard valuation methods are difficult to apply to Anthropic because few listed companies are comparable. Forecasting cash flows is also hard for a new-industry company whose revenue has grown more than tenfold in a year.

“There aren't many similar listed companies to compare to and forecasting the cash flows of a company in a new industry whose revenue has grown more than 10fold in a…”—

AI Market Estimates — Morgan Stanley estimated that generative AI could tap a $60 trillion market; the presenter notes that the firm is likely to underwrite Anthropic’s IPO. Separately, Anthropic’s internal research arm modeled an extreme scenario in which AI adds more than $10 trillion to U.S. GDP by 2030.

“And last week, Morgan Stanley estimated the generative AI could tap $60 trillion, so about half of everything produced on the planet in a year…”—

DATA-CENTERS

Rate Pressure — The presenter links the 10-year Treasury yield reaching 5.23%, its highest level since 2004, to a hot economy, sticky inflation, and a large government deficit. Matt Kennedy says higher rates pressure companies like Anthropic by lowering the present value of distant profits and raising data-center borrowing costs.

“The same survey that showed that the economy is booming sent the 10-year Treasury yield to 5.23% on Thursday evening, the highest it's been since 2004.”—

SB Energy Buildout — The presenter says SB Energy’s hoped-for $50 billion valuation would equal roughly 400 times annual EBITDA. Its prospectus says it needs $174 billion to build what it has already promised, even though it has not yet switched on a data center.

“Now, $50 billion is quite an achievement for any company, but it's especially impressive when you consider that SB Energy hasn't switched on a single data center yet.”—

Depreciation Timing — A data center under construction is classified as construction in progress and is not depreciated; the presenter adds that purchased GPUs are not depreciated until switched on. Depreciation can therefore be delayed while facilities or chips remain unused.

“A data center that's still being built sits on the balance sheet as a construction in progress, and construction in progress doesn't get depreciated.”—

IPO-MARKET

Weak IPO Market — Only three IPOs had occurred since Labor Day, and five of the 10 largest listings that year were trading below their offer prices. WholeTech postponed its IPO, citing market conditions, while SB Energy had filed before Labor Day but had not begun marketing its shares.

“WholeTech, a nuclear power company, with nuclear being the power source the tech industry is counting on to run its data centers, postponed its IPO last week, citing market conditions…”—

PUBLIC-COMPANIES

Nvidia Valuation — Nvidia’s sales have risen from about $27 billion four years ago to an estimated $410 billion this year, with net income expected to almost double. Despite that growth, the presenter says it trades at less than 17 times expected next-year profits, and investors are paying about half as much per dollar of profits as they did last year. He contrasts that valuation with Anthropic’s $2 trillion target.

“Bloomberg reports that its sales have gone from about $27 billion four years ago to an estimated $410 billion this year and its net income is expected to almost double.”—

Investment Gains — Unrealized gains on Amazon’s stakes in Anthropic and OpenAI accounted for 85% of its net income in its latest results. Gains on Google’s stakes in SpaceX and Anthropic accounted for 87%.

“And on the Prof Markets podcast, Ed Ellson pointed out that 85% of Amazon's net income in its latest results came from unrealized gains on its stakes in anthropic and…”—

Anthropic IPO Valuation

The host contrasts Anthropic’s reported $2 trillion target with a strong tech market but weak IPO activity, then explains why its valuation is difficult to ground in conventional forecasts.

Revenue Multiple Reality Check

Using Scott McNeely’s dot-com-era warning and a simplified cash-flow calculation, the host argues that Anthropic’s reported valuation depends overwhelmingly on future growth.

AI Market-Size Claims

The host challenges rapidly expanding AI total-addressable-market estimates, noting that companies often capture little of the markets used to support ambitious valuations.

Data Center Construction Risks

The host questions SB Energy’s valuation before it has operating data centers and describes the financing, construction, depreciation, and equipment-life risks facing the sector.

Circular AI Financing

The host traces how SoftBank, OpenAI, Nvidia, and other firms fund, lease, guarantee, and invest in one another, creating interdependent valuations and risks.

NVIDIA Valuation Puzzle

The host compares Nvidia’s profits with AI labs’ speculative valuations and outlines concerns about peak margins, customer spending, uncertainty, and private-market pricing.

AI Price Cuts and Demand

Cheaper models and falling inference costs threaten AI labs’ pricing power, while rising usage and possible customer lock-in offer a counterargument to the bearish case.

IPO Market Reality Check

The host argues that an IPO would put a public price on private AI stakes, warns of market and issuance risks, and concludes investors should demand evidence behind a $2 trillion valuation.

Related Episodes

Important Disclaimers

Not Investment Advice: For education only — not financial, tax or legal advice.

AI-Generated: Made by AI from source transcripts and may contain errors.

No Guarantees: Past performance does not predict future results; investing carries risk.

Our Commentary · Request Removal