Hyperscalers
Independent fundamental valuations, DCF intrinsic models, and thematic research covering Hyperscalers.
Alphabet (GOOG): Cloud doubles to $30.7B. -$9.1B cashflow and Search squeeze send price down 12%
Alphabet’s Q3 had two opposing forces: Google Cloud doubled revenue with an 80% growth beat and a $561B backlog, but Search slowed as AI answers monetised below classic queries, and $51.4B in capex drove free cash flow to -$9.1B. With the stock down ~11% pre-market to $287, the market is pricing in cash burn and Search cannibalisation while discounting the long-term value of proprietary TPU hardware and Cloud momentum. At ~19x 2027 EPS, the valuation offers upside, but near-term pressure remains until cash flow catches up.
Hyperscalers' two-customer problem: $1 trillion of commitments or who breaks first
Two AI labs, OpenAI and Anthropic, sit inside the backlogs of all four clouds, and the clouds also own stakes in them. Anthropic’s IPO filing shows at least $518B of compute commitments, $252.5B of it with the three clouds, against ~$4.6B of 2025 revenue; OpenAI has committed $250B to Microsoft, $138B to Amazon and about $300B to Oracle. Together the two labs have committed about $940B to the four clouds. Lab-linked contracts are roughly 22% of Alphabet's $514B backlog, 45-50% of Oracle's $664B and Amazon's $496B, and ~46% of Microsoft's $678B. On the equity side, Amazon has committed up to $83B across both labs, Alphabet up to $40B and Microsoft holds ~25% of OpenAI. Cloud growth therefore depends on lab financing as much as on enterprise demand. Alphabet is the least exposed. Oracle, funded by debt, is the most.