AI Chip Bottleneck

Compute demand doubling every 6-9mo · supply chain cannot scale

✦ OPPORTUNITY

Global demand for AI training compute is doubling every 6-9 months. NVIDIA, TSMC, and ASML sit at the center of a hardware supply chain that cannot scale fast enough to meet demand. Waiting lists for H100/H200 clusters extend 12+ months. The constraint is not demand but manufacturing yield and specialized packaging capacity.

Signals (90d)

Opp Countries

Risk Countries

Wagers lens

AI-matched prediction markets — agree or disagree, the decision is yours. Clicking opens the provider's site.

No relevant prediction markets yet. The mapping worker runs daily at 09:00 UTC — check back after collection, or browse all prediction markets.

For entertainment and research purposes only. OpenWatch tracks trends and signals — not real-time prices. Data updates every 4 hours. The forecasting algorithm is currently undergoing back testing, and we do not recommend any position. All trading decisions are solely your responsibility.

Markets are matched to OpenWatch scenarios by an AI worker that runs every 4 hours. New markets and price changes may not be reflected immediately.

Intelligence Evidence — Showing the work

No signals matched yet — check back as new intelligence is collected.

Supply Chain Plays

Who supplies the winners? — second-order beneficiaries often overlooked by direct-theme investors.

TSMThe foundry at the center of every AI chip; 3nm/2nm nodes are the only path to H-series scale LRCXEtch tools are the critical path in advanced logic and 3D NAND; demand scales with AI capex KLACProcess control and metrology; every advanced node requires more inspection steps — KLA sole-sources most of them MUHBM3e memory supply-constrained for H100/H200; each AI cluster needs 80–120GB of HBM

AI Chip Bottleneck

Intelligence Evidence — Showing the work

Supply Chain Plays

Key Companies — Winners & Losers