Here's Why I Believe AI Is Not a Bubble

The article argues the AI rally is not a bubble, highlighting multi-year AI infrastructure demand and chip supply commitments as key support for fundamentals. It cites Micron’s fiscal 2026 Q3 revenue more than quadrupling YoY and notes multi-year customer deals with durability/predictability, while Amazon raised its annual capex forecast from $200B to $220B due to higher memory costs. Overall, it frames “physical AI” (humanoid robots and autonomous vehicles) as an emerging 5–10 year catalyst that could broaden compute and component demand beyond hyperscalers.

Analysis

This is less a pure "AI bubble" call than an argument that the spend is becoming contractual. That matters because multi-year purchase commitments shift the profit pool upstream: memory and other constrained inputs can sustain pricing power and better visibility, while the broad AI basket remains more exposed to narrative multiple expansion than to actual cash flow.

Second-order, higher memory and compute costs are a tax on hyperscaler unit economics. AMZN and MSFT can absorb it near term, but if inference pricing does not rise fast enough, the margin pressure will force rationing elsewhere; that is where smaller AI infra names like IREN are most vulnerable because they lack balance-sheet depth and scale purchasing power.

The contrarian miss is timing: physical AI is a 6-18 month adoption curve, not a next-quarter earnings driver. TSLA’s robotics/autonomy option is real but too far out to be the core trade here; the cleaner signal is still the semicap supply chain. What would break the bullish view is a hyperscaler capex guide cut, or evidence that DRAM/NAND spot pricing and contract terms are normalizing faster than expected.

Key Decisions for Investors

• Buy MU on 2-4% pullbacks; 3-6 month target is +15-20% if contract pricing holds, with a stop if management signals backlog is reverting to spot-like terms or ASPs flatten.
• Pair long MU / short IREN on a 1x1 basis for 1-3 months: quality duration and pricing power versus financing- and power-sensitive AI infrastructure beta. Cover if IREN secures take-or-pay capacity deals or MU guide disappoints.
• Add AMZN only on weakness, not strength; the higher memory-cost environment supports AWS share gains, but the position should be trimmed if AWS operating margin compresses more than 100-150 bps over the next two quarters.
• Do not chase TSLA on this memo. If you want exposure to physical AI, use a small 12-18 month call spread only after concrete robotics/autonomy revenue milestones appear; otherwise the thesis is too far out to justify risk now.
• Short BIRD on any promotional squeeze if borrow is available; this is a liquidity and mean-reversion trade, not a fundamental AI winner.