THE REAL REASON $GOOGL IS GETTING HAMMERED AFTER EARNINGS Google selloff is less a rejection of its AI position than a repricing of economics of replacing Search. Search revenue grew 12% to $63.35B versus 17% in Q2 while query growth reached a five-year high since management had flagged tougher comparisons and currency so slowdown cannot be attributed entirely to AI Mode, and Alphabet doesn't disclose revenue per AI query. The important distinction is incremental versus cannibalized searches. On an illustrative sensitivity, a permanent 5% to 10% shortfall against today's quarterly Search revenue would remove $3.2B to $6.3B in sales and, at an assumed 50% contribution margin, $1.6B to $3.2B of quarterly operating profit. That's the economic risk Cloud now has to offset, not simply the headline Search miss. What makes this more complicated is that Cloud is already offsetting a meaningful amount of that risk. Revenue doubled to $30.7B, while operating profit increased $7.5B to $11.1B, implying a remarkable 48% margin on the incremental $15.6B of sales. Even so, that shouldn't be extrapolated indefinitely because Alphabet now books TPU system sales in Cloud, and those hardware economics may differ from recurring compute services. June's SEC filing reported approximately $514B in Cloud commitments, with just over half of total backlog expected to be recognized over 24 months, but contractual backlog isn't cash or profit, and deliveries depend on capacity and consumption. Google has created an external market for proprietary TPUs, yet we still don't know their unit margins or returns on capital. Cash flow is where the investment debate becomes much less comfortable since Google generated $42.3B of operating cash against $51.4B of capex, producing $9.1B in negative free cash flow adding back the previously accrued $5.2B EU fine still leaves roughly $3.9B negative. Inventory absorbed another $3.2B of cash, some associated with contracted TPU systems, but inventory turning into sales won't eliminate depreciation, electricity or recurring equipment replacement. Equally important Alphabet issued $12.4B of common shares and $28.7B in notes while repurchasing none and shared AI and corporate operating costs rose $3.2B year over year. Its $8.7B unrealized investment loss affected EPS, not operations. The question is whether Cloud's added earnings exceed the lifecycle cost of the infrastructure, not whether revenue keeps doubling. At roughly $3.5T in after-hours market value, an illustrative sum of the parts assigning 25x annualized Cloud operating profit implies about $1.1T for Cloud and leaves $2.4T, or roughly 16x annualized Services operating profit, for the rest, before adjusting for central expenses, other assets and debt. That makes the stock look interesting rather than obviously cheap, especially because Cloud profit doesn't include all shared AI costs. Across the next two to four reports, I want to see Search monetization stabilize as AI Mode gains share, Cloud margins remain resilient despite TPU mix, and operating cash flow begin outpacing capex without further dilution. If those three converge, today's discount may be a significant mispricing. If not, the market is correctly charging Alphabet for a much more capital-intensive future.