The moment the prospectus is laid out, the pricing logic behind Wall Street’s push for a $2 trillion valuation for $ANTHROPIC finally comes into view. With 2025 revenue of $4.59 billion and operating losses of $8.06 billion, what stands out even more is the $518 billion in compute power and cloud infrastructure commitments locked in for the future. This figure is more than twenty times its book cash—essentially converting multiple years of anticipated capital expenditures into inflexible liabilities.
Such extreme balance-sheet expansion pushes the risk exposure of the entire AI arms race to its limits. When nearly a quarter of revenue is supported only by two customers that have not signed any long-term lock-in commitments, any fluctuation in front-end revenue will be magnified multiple times by the back-end inflexible compute-power debt. The market’s current frenzy, in essence, is using an extremely high risk appetite to pre-spend the script’s projected thousands of billions in revenue in 2028.
For the broader liquidity backdrop, this is a stress test of liquidity and risk appetite. If enterprise monetization slows afterward, the enormous compute spending will quickly turn into liabilities that squeeze profits—then feed through to position adjustments in tech-weighted and high-risk assets. The next thing most worth watching is the real retention and compute-fulfillment capacity behind this massive flywheel as it plays out with enterprise customers.
Such extreme balance-sheet expansion pushes the risk exposure of the entire AI arms race to its limits. When nearly a quarter of revenue is supported only by two customers that have not signed any long-term lock-in commitments, any fluctuation in front-end revenue will be magnified multiple times by the back-end inflexible compute-power debt. The market’s current frenzy, in essence, is using an extremely high risk appetite to pre-spend the script’s projected thousands of billions in revenue in 2028.
For the broader liquidity backdrop, this is a stress test of liquidity and risk appetite. If enterprise monetization slows afterward, the enormous compute spending will quickly turn into liabilities that squeeze profits—then feed through to position adjustments in tech-weighted and high-risk assets. The next thing most worth watching is the real retention and compute-fulfillment capacity behind this massive flywheel as it plays out with enterprise customers.