Andy Jassy and Amazon are moving toward a tactic of financial capture in the AI market. The total volume of investment in Anthropic is increasing to $33 billion, but do not be quick to believe in the retail giant's generosity. According to the report, $20 billion of this amount is strictly tied to commercial milestones. In essence, Amazon is turning investments into a tool of control, turning the developer of Claude into a kind of "digital serf."
We see the classic mechanics of round-tripping on an industrial scale: Anthropic has committed to return more than $100 billion to Amazon over 10 years through the lease of AWS capacity. The deal obligates the startup to use Amazon's custom hardware—Graviton processors and Trainium chips (from the second to the fourth generation). As Andy Jassy explained, this ten-year commitment to Trainium is necessary to legitimize Amazon's own silicon in the fight against Nvidia's dominance. By booking 5 GW of capacity for Anthropic, AWS is boosting demand and revenue figures for its cloud, while simultaneously subsidizing the development of its own hardware.
This closed economy deprives the market of a transparent assessment of computing costs. Anthropic justifies the deal with explosive growth—according to their figures, annualized revenue has climbed from around $9 billion at the end of 2025 to a massive $30 billion. However, a trap is hidden in this triumph: a fatal dependence on Amazon's specific architecture limits the mobility of Claude models. If the project's success is inextricably linked to chips that competitors do not have, Anthropic may never be able to change providers.
In our view, this is a dangerous precedent for the formation of an artificial monopoly, where capital circulates within one structure without leaving Amazon's balance sheet. For businesses, this is a worrying signal: the technical independence of AI models is being sacrificed for the scale of computing. You should take into account that AWS is now not just a host for Anthropic, but a full owner of the entire supply chain and the company's operational future. The valuation of AI market leaders is finally breaking away from market realities and becoming a matter of internal accounting for cloud giants.