Musings on Markets (Aswath Damodaran) ·
The Anthropic Feedback Loop: The Cost of Capital Delusion and Diversion!
By Aswath Damodaran
ChorusBring the market's conversation to your AI. Learn more
Summary
Aswath Damodaran values Anthropic at about $1.45 trillion in equity, built on a target of roughly $1 trillion in annual revenue in ten years and a 32.5% target operating margin before a regulatory haircut. He argues Anthropic can keep pricing power and stickiness through Claude agents and client data, while flagging founder control via a public benefit corporation and voting shares as a governance red flag. On balance sheet risk, he distinguishes Anthropic from CoreWeave: Anthropic's debt is mainly contractual compute commitments of about $450 billion, backed by $114 billion in cash, so he builds in a 10% failure risk rather than seeing default as imminent. He also cautions that data center permitting, political backlash and AI regulation could undermine the compute and revenue assumptions, and he argues that high venture and private equity target returns are unrealistic against a market-implied US equity expected return of about 9%.
Key takeaways
- Aswath Damodaran (bullish on Anthropic): Anthropic can reach $1 trillion in annual revenue in ten years, winning premium AI.
- Aswath Damodaran (bearish on Anthropic): Founder control under Anthropic's share structure is a red flag that could prove fatal.
- Aswath Damodaran (bearish on CoreWeave): CoreWeave has abandoned corporate finance first principles and risks bankruptcy from debt-funded data centers.
- Aswath Damodaran (bullish on Anthropic): Anthropic can sustain pricing power via sticky Claude agents, supporting a 32.5% target margin.
- Aswath Damodaran (bullish on Anthropic): Anthropic is no CoreWeave; its debt is mostly spread-out commitments and its cash covers early years.
What was said, by company and theme
9 views from 1 speaker. Every quote is checked against the post's text (how).
Anthropic
Across the shows: 101 speakers in other episodes, 127 bullish and 89 bearish calls, past 30 days
Where everyone stands on Anthropic
Aswath Damodaran · NYU Stern School of BusinessBullish
Anthropic can reach $1 trillion in annual revenue in ten years, winning premium AI.
Target revenue of about $1 trillion in ten years is the input the author says his valuation most revolves around. Revenue would put Anthropic at or near the top of all companies by annual revenue, alongside Amazon and Walmart. The author argues that Anthropic emerges as a winner in the premium portion of the AI market, capturing dominant share with pricing power. Revenue is tied to what AI agents can do and how that demand is monetized. The valuation comes to $1.45 trillion in equity value and depends on this trajectory. If revenue falls short, the author expects lower margins, idle compute and higher failure risk. He calls it the most fragile input but still finds the valuation plausible, while acknowledging he will be wrong on some inputs.
“If I were forced to pick one variable around which my valuation revolves the most, it is my estimate of the target revenues for Anthropic , which I estimate to reach a trillion dollars in ten years.”
Aswath Damodaran · NYU Stern School of BusinessBearish
Founder control under Anthropic's share structure is a red flag that could prove fatal.
The author calls the combination of a public benefit corporation and a voting share structure that keeps founders in control, regardless of equity raised, a red flag. He rejects the view that founder control is a strength because the founders are smart. Anthropic's pathways to riches come with technological, regulatory and competitive setbacks. Managing those requires quick, decisive changes, sometimes abandoning existing policies, and often needs outsiders to challenge management. He says the founders seem to share a common worldview and operate in an echo chamber. His concern as an investor is that they may not see changes in business economics, or may be unable or unwilling to adapt, which he says may be fatal. He admits he does not know the founders and says this may be unfair.
“Even if you accept these views I would still argue, perhaps even more forcefully, that concentrating power in the hands of founders is a bad idea... My concern, as an investor, would be that they will not see changes occurring in business economics... and for Anthropic, that may be fatal.”
Aswath Damodaran · NYU Stern School of BusinessBullish
Anthropic can sustain pricing power via sticky Claude agents, supporting a 32.5% target margin.
The author's profit story is that Anthropic will deliver sustained operating profits, with a target operating margin of 32.5% before the regulatory haircut. He accepts critics' point that AI products may lack moats but does not think the business will be commoditized. He argues that in technology the winners make products stickier, so customers stay even when alternatives are technically superior or cheaper. He says Anthropic has work to do on stickiness but he can see how Claude agents, combined with client data on customers and employees, could become sticky. That would let the company keep pricing power and market share. He says the margin, pricing and unit economics assumptions could each break.
“I can visualize ways in which Claude agents, twinned with client data on customers and employees, can become sticky, allowing the company to preserve its pricing power and market share.”
Aswath Damodaran · NYU Stern School of BusinessBullish
Anthropic is no CoreWeave; its debt is mostly spread-out commitments and its cash covers early years.
The author rejects the view that a money-losing company borrowing hundreds of billions makes the valuation moot. He says Anthropic is no CoreWeave, because most of its debt sits in contractual compute commitments spread over time. Anthropic had a cash balance of $114 billion, which will rise above $200 billion after the IPO, and that is enough to cover its first few years of commitments. Commitments due are about $70 billion next year and $76 billion the year after. He says worries about default in the next one to five years are not in his top ten concerns. He still builds in a 10% failure risk for headwinds and fire sales. He estimates contractual commitments at just over $450 billion, with present value of about $373 billion, against outside estimates of $400 to $800 billion.
“Anthropic is no Coreweave, especially if you consider that most of its debt is in contractual commitments spread out over time and that it will have a cash balance large enough to cover its first few years of those commitments.”
CoreWeave
Across the shows: 79 speakers in other episodes, 23 bullish and 84 bearish calls, past 30 days
Where everyone stands on CoreWeave
Aswath Damodaran · NYU Stern School of BusinessBearish
CoreWeave has abandoned corporate finance first principles and risks bankruptcy from debt-funded data centers.
The author agrees with critics that some AI ecosystem players, naming CoreWeave, have abandoned corporate finance first principles and risk bankruptcy. He says the industry's almost mindless use of debt to fund data centers is senseless for both borrowers and lenders. CoreWeave has become the poster child for this, driving an explosion of concern about default risk and rising market-implied default rates and spreads for companies in the space. The author uses CoreWeave as a contrast with Anthropic, whose debt is mostly contractual commitments and whose cash cushion is large.
“I agree with you on the general argument that some players in the AI ecosystem, such as Coreweave, have abandoned corporate finance first principles and risk bankruptcy”
Venture capital & startups
Where everyone stands on Venture capital & startups
Aswath Damodaran · NYU Stern School of Business
VC and PE target returns of 20-50% are fiction; realised returns are close to index funds.
The author says venture capitalists routinely set target rates of 20%, 30% or 50%, with angel financing highest. Private equity sponsors show investments that on paper deliver IRRs of 20% or more. He says these targets are fiction, because the returns that VCs and PE investors actually deliver are much lower. They are also surprisingly close to what investors can earn on public equities, such as index funds. He suggests these numbers may be marketing props rather than financial metrics. He uses this as evidence that very high required returns are not achievable in practice.
“We know that these target returns are fiction, because the returns that venture capitalists and private equity investors actually deliver are not just much lower but surprisingly close to what investors can earn on public equities, say on index funds.”
US stock market outlook
Where everyone stands on US stock market outlook
Aswath Damodaran · NYU Stern School of Business
A 12-15% cost of equity for US firms is delusional given the market's implied 8.99% return.
The author's market-implied expected return on US equities was 8.99% at the start of October 2026. He says this number is model-agnostic and holds whether or not one believes in CAPM. Pushback on cash flows or growth will lower that expected return, not raise it. In this market, a median US company cannot earn 12% or 15% on an investment of equivalent risk, so such costs of equity border on delusional. He adds that a 15% cost of equity implies a P/E near 7.5, using 5% nominal growth and a 20% return on equity. The actual forward P/E is closer to 20. History points the same way: the equity premium over treasuries has been a tight 7 to 7.5%.
“contending that the cost of equity for a median US company should be 12% or 15% borders on the delusional, since you cannot make that return on an investment of equivalent risk in this market.”
AI infrastructure & data centers
Where everyone stands on AI infrastructure & data centers
Aswath Damodaran · NYU Stern School of Business
Building new data centers is getting harder as local communities and politicians push back.
The author's valuation assumes Anthropic can keep adding compute at lower cost as chip prices decline. He notes that this assumes the compute build-out proceeds smoothly through resource constraints and pushback. He says news stories about data centers make it clear that building new ones is getting more difficult. Local communities are pushing back, and politicians have realised that opposing data centers wins votes. He treats this as a real risk to the compute growth his story needs.
“If you have been tracking news stories about data centers, it seems clear that building new data centers is getting more difficult, as local communities push back, and politicians realize that opposing them is a net vote winner.”
AI regulation & safety
Where everyone stands on AI regulation & safety
Aswath Damodaran · NYU Stern School of Business
A single election could bring onerous AI rules that cut AI revenues and profits sharply.
The author says regulatory risk is as much driven by politics as by economics, and it is the risk he most worries about being wrong on. His base case assumes regulation costs only a small slice of operating profits. But a major election could bring onerous AI regulation that takes a major bite out of revenues and profits. He ranks regulatory risk lower than his main inputs because it is driven by operating variables. He still says it is the area where his view is most exposed to error.
“You could be one major election away from onerous regulations on AI, where unlike my base case, where those regulations cost you only a small slice of your operating profits, they take a major bite out of revenues and profits.”
Across the shows
What other episodes Chorus covered said about the same companies in the 30 days up to this one.
Topics
- AI regulation risk
- balance sheet and default risk
- data center siting and permits
- expected returns and cost of equity
- governance structure
- neocloud debt and bankruptcy risk
- pricing power and stickiness
- revenue trajectory
- target vs realized returns
More from Musings on Markets (Aswath Damodaran)
- The Cost of Human Frailty: Nvidia's Jensen Huang Test! · November 19, 2026
- The Rocky Road from Aggregate to Per Share Value: Anthropic's IPO update · November 19, 2026
- The AI Trade Deconstructed: AI Builders, AI Suppliers and AI Product firms! · November 9, 2026
- The Anthropic Prospectus Arrives: An IPO valuation! · October 28, 2026
Ask your AI about this post
Chorus gives Claude, ChatGPT and other assistants every view from this post and every other show Chorus follows. Ask who agrees, who changed their mind, and what else was said about the same companies.
Chorus reports what speakers said. It isn't investment advice. How Chorus works · About Chorus