AI Bottlenecks Dashboard
What it is
Public dashboard at https://aibottlenecks.app/app that maps AI infrastructure into physical bottleneck baskets. The dashboard expands the original ai bottleneck beta article from a narrative claim into a live watchlist, theme basket, and catalyst tracker.
Observed structure
- Watchlist: 96 names.
- Theme baskets: labeled 15 baskets; observed themes include InP & Substrates, Photonics / CPO, HBM / Packaging, Memory Supercycle, Power & Grid, Cooling, Custom Silicon, Networking / Retimers, Connectors & Interconnect, AI Cloud / Neoclouds, Construction & MEP, Storage, Rare Earths, and Lithography & Fab Tools.
- Catalyst tracker: observed next catalyst was ALAB Q1 print plus COHR Q3 FY26 print; China Ga/Ge/Sb expiry was shown as Nov 27, 2026.
- Quote source note: Yahoo ~15m delay.
What the dashboard adds to the thesis
The dashboard gives ai physical stack watchlist a concrete implementation: a 96-name universe, theme groupings, conviction labels, target-weight points, bull/bear targets, triggers, and equal-weight basket charts. It also strengthens the route-card framing behind hbm manufacturing bottlenecks by separating the AI trade into manufacturing and infrastructure stations rather than a single broad AI bucket.
Data-quality flags
- Basket methodology is not fully documented in the observed page text.
- Basket counts in the theme tab differ from watchlist counts for several categories, so the performance charts may be based on selected subsets rather than every watchlist name.
- The theme tab says 15 baskets, but the watchlist themes extracted from the client bundle total 14; the visible theme tab also includes an Outliers basket with 0 names.
- Displayed returns are dashboard outputs, not independently verified market data.
- Bull/bear targets are useful for monitoring but require primary-source valuation work before investment use.
Diligence priority
Before using the dashboard for position sizing or basket construction, verify:
- Constituent inclusion rules.
- Equal-weight methodology and rebalance rules.
- Price source, split/dividend adjustment, and currency conversion.
- Whether returns are point-in-time or subject to survivorship / look-ahead bias.
- Material revenue exposure of each company to its assigned bottleneck.
Related pages
Glass substrate follow-up
The dashboard already flagged glass/T-glass as a next-layer watch item; PhotonCap provides a more explicit framework for the glass substrate cycle, including TGV/metallization bottlenecks and a GVM scoring framework. See glass substrate cycle.
Local source refs
raw/articles/ai-bottlenecks-dashboard-snapshot-2026-05-11.mdraw/articles/photoncap-glass-substrate-cycle-2026-05-08.md