Research Firm Calls Anthropic IPO Valuation Absurd, Sets Target at Only $15 Billion
nashnova research
Independent research firm New Constructs has labeled Anthropic's upcoming IPO "the most ridiculous IPO of 2026," assigning a $15 billion target valuation — less than 1% of the company's roughly $2 trillion listing target — and questioning whether any closed-source AI model can sustain a viable business.
$15 billion vs $2 trillion — where does the gap come from?
New Constructs argues that to justify a $2 trillion market cap, Anthropic would need to earn twice Nvidia's net income over the past four quarters — and Nvidia posted over $190 billion in net profit during that span.
This means → the market is pricing a company that lost $42 billion last year as if it will soon out-earn today's most profitable chipmaker by a factor of two.
In plain terms = buying at this price is not a bet that Anthropic turns profitable next year — it is a bet that it becomes twice as profitable as Nvidia is right now.
How much is Anthropic actually making — and losing?
According to Reuters, citing a leaked prospectus, Anthropic recorded $4.6 billion in revenue in 2025 alongside a net loss of $42 billion.
The company itself disclosed in July that its annualized revenue run rate had grown sevenfold to $65 billion; the New York Times reported in September that annualized revenue could reach $100 billion by the end of 2026.
This means → revenue growth is genuinely staggering, but losses are scaling just as fast — income is running, and cash burn is running faster.
Open-source models are rising — can closed-source still make money?
New Constructs' core thesis: the rise of open-source models casts doubt on the profitability of closed-source ones. The report states plainly, "We believe Anthropic has no viable business model."
In plain terms = when competitors release comparable models for free, charging for your model gets harder with each cycle — that is the structural pressure every closed-source AI company faces.
Anthropic itself flagged in its prospectus that AI could pose "catastrophic or existential risk" to humanity; New Constructs cites this as another reason investors should steer clear.
Does this research firm have a track record worth trusting?
New Constructs founder David Trainer is known for bearish IPO calls. In 2019 he called WeWork "the most ridiculous IPO of the year" — WeWork later pulled its listing and filed for bankruptcy in 2023. He was right that time.
But in 2020 he compared DoorDash to WeWork, calling it "similarly disadvantaged." DoorDash debuted at over $60 billion and now trades at roughly $83 billion. Trainer himself conceded, "Strange things happen."
This reflects a mixed prediction record — on an extreme-bear stance, the firm has scored clear hits and clear misses in roughly equal measure.
What is this IPO really for?
The report argues bluntly: the real purpose of this IPO is not to create wealth for public-market investors but to provide exit liquidity for the Wall Street backers behind the company.
This means → in New Constructs' view, early investors are the primary beneficiaries of this listing; retail investors look more like "the exit."
A key caveat: New Constructs has not seen a formally published prospectus — its analysis relies entirely on leaked data reported by media outlets. Whether the $2 trillion valuation holds up in the public market remains the ultimate test.
市场有风险，内容仅供研究参考，不构成投资建议。