XNews

Apple and Amazon AI Earnings Analysis

Apple and Amazon AI Earnings Analysis

Stratechery analysis of Apple and Amazon earnings reveals AI is reallocating computing resources, squeezing Apple’s supply while boosting AWS. - Apple’s iPhone revenue grew 22% and Mac 29%, but shares fell 7% post-earnings due to a supply-side miss: September quarter guidance of 9%-11% revenue growth was below expectations. - Tim Cook blamed demand forecasting, not typical supply issues; Apple’s historical flexibility on TSMC’s most advanced nodes has vanished as Qualcomm, AMD, Microsoft, and Amazon compete for N3P capacity. - Apple prepulled future N3 wafer starts to meet current demand, creating a future window where N3 production declines; iPhone 17 Pro is expected to sell out and M5 series chips to face severe shortages. - Apple, once seen as an AI-expenditure safe haven, is indirectly exposed to AI-driven chip supply competition. - Amazon: AI is pulling AWS core services upward; Jassy noted post-training reinforcement learning and agent workloads run on CPUs where Graviton has 30%-40% better price-performance, and AI data drives database/storage usage. - Jassy stated “no one model will rule the world” but confirmed Amazon is still building frontier models for cost, priority, and speed; AWS positions as neutral Bedrock host. - AI adoption is dumbbell-shaped: AI labs and a few apps on one end, enterprise automation on the other, with massive mid-range enterprise workloads yet to migrate to AI. - Trainium chip business is over $25B annualized, currently internal but with growing external demand, similar to Google’s TPU trajectory.

Key insights

Apple’s Lost Privilege: TSMC Advanced Capacity Shared Among AI Chip Competitors

For years, Apple effectively used TSMC’s most advanced lines as its buffer; now N3P is shared with Qualcomm, AMD, Microsoft, and Amazon. This loss of supply elasticity led to a demand forecast miss and will constrain iPhone 17 Pro and M5 availability. Apple’s stock drop reflects the market’s realization that even non-AI companies face AI-inflicted supply chain risk.

AWS Benefits from AI in Multiple Layers, Not Just Accelerator Rental

AI drives CPU usage for reinforcement learning and agent tool calling, where Graviton’s cost advantage matters. AI data also increases demand for storage, databases, and vector indexes. This multi-faceted pull makes AWS a core beneficiary even if it doesn’t lead in frontier model training.

Dumbbell AI Adoption Curve: The Enterprise Middle Is the Ultimate Prize

Current AI compute is concentrated at the extremes—research labs and narrow enterprise wins. The largest opportunity is migrating mainstream enterprise workloads to AI inference, which Jassy predicts will make AWS a trillion-dollar business. This middle won’t be converted by persuasion but may be swallowed by AI-native challengers or cost-driven automation.

Amazon’s Custom Silicon (Trainium) Is Becoming a Cloud-Independent Revenue Stream

Trainium chips are currently internal but generate over $25B annualized revenue, with interest from customers like Meta wanting direct chip access. This mirrors Google’s TPU strategy and signals that custom AI silicon will be sold outside the cloud, altering the chip market.

Source posts