AI power: rack and chip conversion
AI power: rack and chip conversion
Rising datacenter power density will increase the commercial content of rack-to-chip conversion and cooling suppliers, provided design wins become revenue without eroding margins.
Watch for: MPWR Vera Rubin socket share + ASP follow-through after positive Q1 catalyst resolution, Vicor Gen 5 VPD ramp 2H 2026, BESIY orders for SK hynix HBM + foundry BSPDN, Intel 18A yield + Panther Lake unit ramp Q2 2026, TSMC A16 SPR volume timing (now 2027, was 2H 2026), NVIDIA Feynman 2028 (first AI accelerator on A16+SPR), Delta Electronics market cap relative to Foxconn, Wolfspeed substrate share plus gross-margin repair (bottoming or terminal? equity re-rate is not enough), MLCC capacity allocation pricing (Murata, Samsung Electro-Mechanics), ETN/Schneider response signal — defensive product announcements OR hyperscaler design-out language
The thesis basket — the 5 tickers this perspective actually tracks. Everything else it has touched is folded below.
Seen alongside in scans (43 shown of 197) — symbols that appear with this perspective, not the thesis basket
Show all 197 ticker links
no live data (1) — unresolved, delisted, or non-US symbols
The Thread
The thesis as it unfolds — by source type (the symbol grid above is the current state).
Recent Shifts
2Signals
1The AI Money Went Into the Ecosystem, Not Into Nvidia
Market signal state changed.
Research
15Captures
12Calendar & Lifecycle
26Frontier Map — the 130-week bushing, not the transformer core, is the grid's longest lead time
Frontier Map — the 130-week bushing, not the transformer core, is the grid's longest lead time
CV Research via CK Capital — AI power loss scales with current squared, not watts, and one vendor transforms instead of switching
CV Research via CK Capital — AI power loss scales with current squared, not watts, and one vendor transforms instead of switching
Bloomberg — Chicago joins the data-center moratorium push
Bloomberg — Chicago joins the data-center moratorium push
SemiAnalysis — AMD Helios vs NVIDIA Vera Rubin: double-wide racks and where the rack-level constraints bite
SemiAnalysis — AMD Helios vs NVIDIA Vera Rubin: double-wide racks and where the rack-level constraints bite