Anthropic IPO leak flags $100B raise and $2T valuation target

If the Softunis and related reporting is accurate, Anthropic is preparing to test whether public markets will underwrite a $2 trillion AI pure play with half a trillion dollars of locked in compute commitments. That is unprecedented capital intensity even by the standards of cloud and chip companies. It signals that the company, and its backers at Amazon, Google, Microsoft, Broadcom, AMD and xAI, are betting that access to frontier scale compute will be the ultimate moat in the run up to AGI.

From a strategic standpoint, the numbers described in the draft prospectus would hard wire Anthropic into the global semiconductor and cloud supply chain for a decade. $518 billion in contractual obligations means foundries, GPU vendors and data center operators can plan around Anthropic’s demand almost as if it were a sovereign buyer. That tilts the playing field further toward a handful of well financed labs and makes it much harder for smaller players to secure comparable capacity, no matter how strong their research.

At the same time, public investors may balk at subsidizing such extreme negative cash flows, particularly when the revenue base is still in the mid tens of billions. If the IPO prices below expectations or trades poorly, it could cool the entire AI capital cycle and force labs to prioritize monetization and efficiency over unconstrained scaling, with second order effects on how quickly AGI class systems are pursued.