AI investing has a pattern everyone ignores: $NVDA and Jensen literally telegraph the next supply-chain bottleneck, and the market dismisses it until it's too late.

2025 playbook: Nvidia locks up EML and laser capacity. Market calls photonics a bubble. A year later: $LITE +678%, $AAOI +475%, $COHR +261%, $AXTI +3,844%.

2026 playbook: Nvidia secures CW/EML capacity with long-term agreements, pushes 800V and CPO architectures, flags Physical AI as the next theme. Market today: "CW names are meme stocks, 800V and CPO are years away, humanoids aren't profitable."

The upstream bottleneck names get written off as hype until Nvidia's capex actually flows through. If you believe Jensen is the leading indicator — and the 2025 returns suggest he is — then the 2026 supply-chain bets (CW lasers, 800V power, CPO packaging, Physical AI enablers) are the 2027 setup. The market learns this lesson every cycle and forgets it just as fast.