Half of Google's and Amazon's 'blowout AI profits’ came from a stake in Anthropic—not from their actual business

Alphabet and Amazon reported massive Q1 2026 profits boosted by unrealized gains on Anthropic stakes, including Alphabet's $36.9 billion equity gain and Amazon's $16.8 billion pre-tax gain from Anthropic investments. The article argues that a substantial share of Big Tech earnings is being inflated by markups on private AI holdings rather than core operating results, even as the companies are spending $130.65 billion on capex in the quarter and plan nearly $700 billion for the year. This is a meaningful earnings-quality story for large-cap tech, but the tone is mixed because headline profits remain strong while their composition looks increasingly volatile.

Analysis

The core market issue is not whether AI capex is large; it is that reported earnings are becoming partially endogenous to the same ecosystem spending that is supposed to justify the capex. That creates a reflexive loop where incremental cloud commitments and investment rounds can manufacture accounting profit before any durable operating cash flow arrives, which mechanically lowers near-term valuation multiples for the platform names while obscuring true economic return on invested capital.

This is structurally supportive for the hyperscalers in the short term because headline EPS reduces the urgency of multiple compression, but it is also a governance and quality-of-earnings risk. If investors begin to discount equity-method gains from private AI stakes, the market will re-rate the segment toward operating cash flow and capex intensity, which is a harder test for AMZN than GOOGL because AWS is already funding a heavy reinvestment cycle. The second-order beneficiary is Anthropic’s competing capital stack: every incremental strategic check from a hyperscaler improves its funding access while also inflating the mark-to-market of the incumbents’ stakes.

The contrarian read is that the market may be underpricing a medium-term accounting reset. If private AI valuations stop compounding at the same pace or if any funding round clears at a lower step-up, reported profit growth could decelerate sharply even if core businesses remain intact. That makes the next 1-3 quarters a catalyst window: the shares can stay supported as long as capex and funding rounds remain bid, but the setup becomes fragile if investors shift from EPS to free-cash-flow yield and capex efficiency as the dominant metric.

Key Decisions for Investors

• Short-dated hedge: buy 1-3 month GOOGL and AMZN put spreads into the next earnings cycle. Thesis: headline earnings quality becomes a focal point; limited downside if the market keeps rewarding AI optionality, but convex payoff if investors start stripping out equity-markup gains.
• Pair trade: long MSFT / short AMZN over 3-6 months. Microsoft’s AI monetization is more directly tied to recurring software usage, while Amazon’s reported profit is more exposed to non-operating markups and capex intensity; risk is that AWS growth re-accelerates enough to offset the accounting skepticism.
• Event-driven long GOOGL vs short a basket of high-multiple AI infrastructure suppliers on any post-earnings rally. If the market starts discounting private-markup earnings, the hyperscalers may underperform less than the supply chain because the latter has less balance-sheet support and more valuation dependence on capex continuation.
• For existing longs in GOOGL/AMZN, trim 20-30% into strength and replace with call spreads rather than outright stock. This preserves upside if AI monetization improves while capping exposure to an earnings-quality reset over the next 1-2 quarters.
• Watch for a downgrade trigger if Anthropic/peer funding rounds come in below prior step-ups or if management starts emphasizing free cash flow over EPS. That would be the signal to rotate from beta-long to neutral/short on the hyperscaler complex.