Living public-equity thesis
AI Supercycle
AI is a capital-spending supercycle whose value capture migrates across bottlenecks. The durable edge is not predicting one permanent winner; it is tracking which layer is constrained, whether utilization and customer economics justify the next round of capacity, and when new supply turns scarcity into commoditization.
Market close · Jul 17, 2026
Thesis v0.1 · 2026-07-17
Thesis conditions and falsifiers
What must hold
- — Training and inference demand must grow fast enough to absorb continuing gains in hardware and model efficiency.
- — At least one physical or silicon bottleneck must remain scarce enough to support pricing, utilization, or backlog visibility.
- — Customers must translate deployed compute into productivity, revenue, strategic advantage, or defensible option value.
- — Capital providers must continue funding long-duration infrastructure before the full application revenue pool is visible.
What breaks it
- — Utilization or order visibility weakens across multiple layers while committed capacity continues to arrive.
- — Model efficiency and open alternatives reduce compute intensity faster than new workloads expand it.
- — Power, permitting, financing, or customer concentration leaves material infrastructure stranded.
- — Application economics fail to support the cost of the underlying compute stack over a full investment cycle.
Sector monitor
Where is the AI trade working?
Equal-weight local price returns across the listed names in each coverage basket. Select a row to open its thesis, chart, ETFs, and members.
| Sector basket | 60D trend | |||||
|---|---|---|---|---|---|---|
| -2.8% | -9.1% | -1.0% | +24.6% | +79.0% | ||
| -1.6% | -7.5% | -1.9% | +37.3% | +63.0% | ||
| -0.3% | -5.5% | -2.4% | +0.7% | +52.0% | ||
| -2.1% | -2.8% | -3.1% | +23.9% | +63.3% | ||
| -2.0% | -7.8% | -3.4% | +27.8% | +52.7% | ||
| -1.7% | -9.5% | -7.7% | +39.5% | +95.1% | ||
| +0.1% | -12.1% | -14.4% | -3.9% | +44.9% | ||
| -9.0% | -14.9% | -24.6% | +22.1% | +23.3% | ||
| -8.3% | -21.4% | -33.5% | +97.4% | +323.7% | ||
| +1.9% | -18.2% | -34.9% | -17.5% | +51.9% |
Price returns only; distributions excluded. Missing exchange sessions are not treated as zero.
Source: Yahoo Finance chart data ↗Selected sector
Semiconductors
Basket close · Jul 17, 2026
Basket vs SPY · rebased to 0%
Coverage basket
Names underneath the thesis
Equal weighted · native-currency returns · 3/3 available
| Name | Venue | 1M | 3M | YTD |
|---|---|---|---|---|
| NVIDIANVDA · NVDA | Nasdaq | -2.2% | +2.2% | +8.7% |
| AMDAMD · AMD | Nasdaq | -2.3% | +78.2% | +131.5% |
| TSMCTSM / 2330 · TSM | NYSE / Taiwan | -6.4% | +9.6% | +31.1% |
Company links open the stored primary source supporting the coverage classification. Performance is descriptive, not a buy or sell recommendation.