Skip to content

Leadership1 publisher3 min readPublished

Investors Are Pricing Anthropic Like Infrastructure. Check Your Exit Clauses.

Serious money is modeling a $2 trillion IPO for a five-year-old model vendor. That is a vendor-concentration problem before it is a bubble question.

The Board Room · Leadership desk

Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction

Illustration accompanying Investors Are Pricing Anthropic Like Infrastructure. Check Your Exit Clauses.
Generated illustration

What happened

  • Investors are modeling an IPO valuation above $2 trillion for Anthropic, the five-year-old maker of Claude, according to the Financial Times.
  • Some investor scenarios for an Anthropic IPO stretch as high as $3 trillion.
  • Anthropic has not publicly set that valuation.
  • A $2 trillion Anthropic would signal that investors expect foundation model companies to become core infrastructure for global business, commanding economic power on the scale of today's largest technology and energy companies, affecting how corporations think about software budgets, automation, vendor dependence and model selection.
  • Anthropic was valued at about $380 billion in February.

Compiled by The Board RoomSomething wrong?How this is made

Why it matters

Investors are modeling an IPO valuation above $2 trillion for Anthropic, the five-year-old maker of Claude, according to the Financial Times as reported by Forbes, with some scenarios stretching to $3 trillion [1][2]. Anthropic has not publicly set that valuation [3], but the number matters to buyers regardless: a price like that says the public market expects foundation model companies to become core infrastructure for global business, on the economic scale of today's largest technology and energy firms [4].

The repricing has been fast. Anthropic was valued at about $380 billion in February and at $965 billion in a May financing round, with secondary market demand pushing implied prices higher since [5][6][7]. A $2 trillion debut would be roughly 5.3 times the February mark [1].

The case for it is growth. Investors cited by the Financial Times expect annualized revenue of roughly $100 billion to $120 billion by the end of 2026, after starting the year at a fraction of that [8]. At $100 billion, $2 trillion is about 20 times sales [9]; the $3 trillion scenario against $120 billion is about 25 times [2]. Those are demanding multiples but not unprecedented ones.

What is unprecedented is the cost structure sitting underneath them. Classic enterprise software earned premium multiples because each additional customer cost little to serve [10]. Frontier AI inverts that: usage creates revenue and usage creates substantial cost, while better models require another generation of expensive hardware [11]. A modern 100 megawatt AI data center can cost more than $4 billion to build and operate, with roughly 70% going to servers and graphics processors, according to a Reuters Breakingviews analysis [12][13] - about $2.8 billion of silicon per site [3]. Apollo and Blackstone are backing a $35 billion capacity expansion tied to Anthropic and Broadcom technology [14], the equivalent of nearly nine such builds [4]. Nvidia, separately, is assembling financing structures intended to support more than $500 billion of compute investment [15].

For comparison, Facebook went public in 2012 at roughly $104 billion, already carrying hundreds of millions of users and a highly scalable advertising machine [16][17]. A $2 trillion Anthropic would be about 19 times that [5]. Saudi Aramco's 2019 IPO valued the company near $1.7 trillion and raised $25.6 billion [18].

The operator's problem is not whether the multiple is right. It is that the market is pricing lock-in that has not been earned. Infrastructure multiples assume switching is hard and pricing power is durable, yet Forbes notes frontier AI companies face aggressive price competition, open source challengers and third party models that can erode their grip on customers [19]. The unresolved question, in the same piece, is whether soaring revenue can produce equally compelling free cash flow [20]. Meanwhile the numbers in circulation are revenue and valuation, not margin. If the vendor's compute bill becomes structural, it becomes your line item, and the $35 billion and $500 billion financing vehicles are the shape of who eventually pays for it [14][15].

Watch for an S-1, which would be the first document to put gross margin, compute commitments and customer concentration on the record. Watch whether the $100 billion to $120 billion run rate arrives on schedule [8]. And before renewal season, audit your own agreements for model portability, price escalators and take-or-pay minimums, because those are the clauses a $2 trillion valuation is implicitly counting on.

Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories