AI power: rack and chip conversion · Campaigns

Campaign

AI power: rack and chip conversion

active · AI · Power

Horizon 6–24 months; test socket conversion and margins at the next two quarterly reports. Last research review Latest evidence Priority high

Notebook

Working notes, questions and changes. Rough material stays unverified until reviewed.

  1. notebookunverified

    Campaign scope review: a smaller working set

    Evidence as of 2026-09-06

    Keep distinct from generation; replace the unverified percentage-loss headline with commercial conversion tests.

    Read working note

    Campaign scope review: a smaller working set

    Reviewed September 6, 2026. Market-price context remains the September 4 settled corpus. Primary disclosures retain their original dates. This review changes research scope and attention; it issues no desk calls and changes no historical grades.

    The live set falls from 22 to 10: five active cases and five monitoring cases. Twelve former live cases close with explicit reasons and conditions for a new successor. The six previously closed records remain closed, with their original closure metadata preserved. Their scope review is archival only; it does not reconstruct missing historical claims or validate their old outcomes.

    What stays live

    Campaign Status Why it earns a separate case Next review
    AI power: generation and grid equipment active Keep the evidenced upstream electricity claim; remove cloud hosts and downstream sockets from its curated instruments. 2026-10-06; earlier on its named condition.
    AI power: rack and chip conversion active Keep distinct from generation; replace the unverified percentage-loss headline with commercial conversion tests. 2026-10-06; earlier on its named condition.
    AI optics: bandwidth becomes supplier earnings active Promote to active research using current supplier evidence; keep the commercial optical case separate from power and memory. 2026-09-30; earlier on its named condition.
    AI memory: pricing discipline through the supply ramp active Keep the memory claim; remove STX/WDC and HDD wrappers so nearline storage owns the HDD economics. 2026-09-30; earlier on its named condition.
    Nearline storage: density and supplier discipline active Keep a separate HDD campaign with current operating evidence; remove its duplicate coverage from memory. 2026-10-06; earlier on its named condition.
    Energy disruption: physical flows and exporter economics monitoring Consolidate energy monitoring around physical flows; retire political dominance and war-ending framing, retaining a conditional evidence watch. 2026-09-09; earlier on its named condition.
    Nuclear fuel: contracting reaches enrichment and supply monitoring Retain the distinct fuel mechanism on monitoring; remove utility and reactor-development overlap. 2026-10-06; earlier on its named condition.
    Biotech funding: conversion into tools and CRO orders monitoring Keep the funding transmission test on monitoring; weed out the unrelated AI-platform and IPO-halo branches. 2026-10-06; earlier on its named condition.
    AI spending: the supplier earnings reversal case monitoring Replace the vague crowded-trade lens with a falsifiable opposing case; monitoring records uncertainty, not a bearish desk call. 2026-09-11; earlier on its named condition.
    Drone defense: contracts become profitable deliveries monitoring Keep a narrow monitoring campaign with a near-term evidence event; remove the broad speculative defense basket. 2026-09-10; earlier on its named condition.

    What leaves the working set

    Closure below means that the case no longer earns standing research attention. It does not mean the market proved it false. Unresolved evidence and historical calls retain their original status.

    Former campaign Disposition Evidence that would justify a successor
    Gulf Infrastructure Strike Scenario Energy-flow context retained; retire the unsupported helium cascade as a live campaign. Direct Qatar operator evidence plus measured helium allocation or fabrication impact would justify a new, narrowly scoped successor.
    Risk-On Tech Rotation — Concentrated Mega-Cap + AI Supercycle Retire the war-ending premise and the generic technology-rotation remainder. The user rejects continued attention to that framing; the existing review also lacks an independent resolution catalyst. Only a new independent market claim with direct evidence warrants a successor. A peace headline alone does not restore this campaign.
    The Gold Anomaly — Broken Safe Haven Retire a price anomaly that has not established safe-haven attribution or a separate investable mechanism; preserve the historical observations. A defined event study separating dollar, rates and risk effects plus an actionable claim could support a successor.
    SpaceX IPO Liquidity Event — $40-80B Drained from Markets, June 12 2026 Close the historical event case as ungraded, not successful or disproven. Preserve its original identity and recovery questions; do not relabel it as another issuer. Recover original filed terms and contemporaneous event-window observations for a retrospective only.
    Livestock Disease — New World Screwworm → Cattle-Supply Shock & Protein Rotation Remove the conditional disease scenario from the working set because official detection and commercial transmission were unresolved in the reviewed packet. An official APHIS detection plus identified commercial animal-health demand or protein substitution would justify a focused successor.
    CPU Shortage Supply Chain — Datacenter CPU Allocation + ABF Substrate Bottleneck Retire the broad CPU-shortage basket until CPU allocation and substrate utilization establish a distinct constraint. MPS power growth is retained in power delivery, not promoted into proof of a CPU shortage. Primary CPU allocation/pricing evidence plus ABF or OSAT utilization and a named beneficiary would justify a separate successor.
    China Structural Discount — Multiple-Capped, Not Earnings-Capped (VIE / CCP / Macro Triptych) Retire descriptive geography coverage without a current rerating mechanism; retain its governance, earnings and wrapper questions for future research. A specific governance or regulatory change with an earnings-versus-multiple study and a dated decision window would justify a successor.
    Physical AI / Robotics — Arms-Dealers Over Moonshots (the Embodied-AI Rotation) Retire the broad robotics basket while scarce component content and profitable robot-specific conversion remain unverified. General machine-vision growth is insufficient. A named component supplier with disclosed robot-program orders and profitable revenue conversion would justify a narrow successor.
    Quantum Computing — The Sovereign-Catalyst Trade (Basket-First, Lottery Underneath) Retire the sector option while funding, runway and repeatable commercial demand remain unresolved; technical or policy milestones alone do not earn a live campaign. Filed funding terms and repeatable paid customer demand with a specific economic mechanism would justify a successor.
    AI Mega-IPO Liquidity Regime — Index Plumbing, Lockups, and the Issuance Bell Retire the generic issuance lens as a standing campaign until a filed event creates a dated decision. Keep the SpaceX retrospective distinct rather than manufacturing a larger live replacement. An actual filing with reconciled float, lockup and index dates, plus a measurable funding-flow hypothesis, would justify a new event campaign.
    Trump Policy Ticker Catalysts — Source-Tier Event Study Retire the general policy watch as a campaign. Direct statements remain research inputs; each concrete economic claim must earn its own case. A direct primary policy action with a specific economic transmission and a response surviving earnings and sector confounds would justify a successor.
    The Midterm Window — Q4 Turn Watch Retire the calendar overlay as an independent campaign. Its breadth and volatility observations belong in the climate review; seasonality alone does not warrant an entry case. A distinct evidence-backed election mechanism beyond the calendar could justify a successor; ordinary breadth changes go to climate review.

    Themes and distinct mechanisms

    Themes group related work without creating another campaign hierarchy. Upstream electricity, rack-to-chip conversion, optics, semiconductor memory and HDD storage have separate commercial tests. The memory instrument list no longer duplicates HDD suppliers. Nuclear fuel excludes the utility-power economics covered upstream. Biotech focuses on tools and CRO orders; drone defense focuses on funded procurement conversion.

    The AI-spending counter-case opposes the optical and memory cases only on their common test through March 2027: sustained supplier pricing and margins versus a spending-driven reversal. It remains monitoring because current supplier growth challenges an immediate-collapse story. The same source can support one campaign and challenge another. Long-lived power projects are related, not automatically opposite.

    The Gulf scenario's physical energy-flow monitoring moves to the retained energy case. Its unverified helium/fabrication branch is retired. Both IPO entries close: the SpaceX event remains an ungraded retrospective, and a future issuer must bring its own filed event and decision window. No broad live IPO replacement is manufactured.

    The user's rejection of the war-ending framing is an attention decision, not evidence about military outcomes or market indifference. The independent reason for closure is the absence of a distinct supported decision beyond generic technology leadership. Energy monitoring uses physical flows, supply recovery and company transmission instead.

    Evidence boundaries

    The clean September 6 thesis-review capsules supplied the existing models, tests and unresolved gaps for all 22 live cases. Raw legacy perspective prose and all account/holding records were excluded. Twelve primary-source assessments were checked and captured for this scope review. New horizons and failure tests are explicit desk hypotheses, not issuer forecasts or trading instructions. A reviewed scope means the question and evidence boundary have been reviewed; it does not mark every underlying assertion as proven.

    The active list is an attention priority. Individual entries still require a separate numeric desk call and company-specific evidence. Instruments removed from a campaign's curated scope remain available in historical events and calls.

    Desk Review 2 trigger check

    No new decision window was named. This remains the same weekend as Desk Review 1. The settled benchmark observations have not advanced beyond September 4; the second desk review's 30-day grades mature October 6 and its 90-day trigger falls December 5. The reviewed primary disclosures predate Desk Review 1; reopening them now is not a new market event. A targeted current primary-news check did not establish a new focus-lane event after that review; it was not a full news acquisition pass. No newly established trigger supports firing the whole desk again today.

    The next manual climate checkpoint remains September 11, earlier on the climate note's existing conditions. The user has authorized Desk Review 2 when a trigger fires. This scope review is not counted as that run, and no automation or scheduler is added.

    Acquisition receipt

    Layer This review
    Settled prices and scan corpus Carried September 4; benchmark date/price comparison performed.
    Primary issuer and official web sources Ran targeted checks; source dates and limitations preserved below.
    Full news/social acquisition Not run; this was a scope review.
    Fresh provider fundamentals or OHLC Not run; no new price levels, financial tables or calls issued.
    Routine due loops Cadence reports five overdue loops: persona pass, curated sweep, weekly money loop, nightly heartbeat and paper cycle. None is claimed complete by this review.

    Sources

  2. notebookunverified

    EOW Phase 0a refresh: the hardware-vs-software rotation goes macro; Broadcom's own call reinforces upstream demand even as its stock fell on the print

  3. notebookunverified

    EOD Phase 0a refresh: NVIDIA's print reinforces upstream demand again; the equipment layer's rotation-map read cools further across a wider adjacent cluster

  4. notebookunverified

    EOW Phase 0a refresh: the hardware layer beneath the cloud is where money rotated out this week; the equipment leg upstream mostly held its long trend

  5. notebookunverified

    EOW Phase 0a refresh: AMAT's checkpoint print lands on the calendar, unconfirmed in the desk's own sources

  6. notebookunverified

    EOD Phase 0a refresh: upstream compute demand strengthens; direct delivery-side proof remains pending

  7. notebookunverified

    EOW Phase 0a refresh: POWI's print lands as an unfilled stub against its own reference-design watch; ZEISS capacity build corroborates the equipment leg

  8. notebookunverified

    EOW Phase 0a refresh: VRT's 10-Q drop is a name-level event; MSFT print reads as upstream demand confirmation

  9. notebookunverified

    EOD full-scan Phase 0a refresh: parent's verdict week reads as upstream demand confirmation, not a direct signal here

  10. notebookunverified

    EOW Phase 0a refresh: parent's GOOGL Q2 capex print reads as upstream demand confirmation

  11. notebookunverified

    EOW Phase 0a refresh: POWI lands a named NVIDIA reference-design win; Helios/Vera Rubin rack interview names storage as a new co-constraint

  12. notebookunverified

    EOW Phase 0a check: no fresh delivery-cohort tape; sector rotation still net-negative

  13. notebookunverified

    AMAT cools from cycle leader to "don't chase"; ON flush continues; semicap/foundry core unchanged

  14. notebookunverified

    EOW positioning flush materialized; MPWR approaching oversold; AMAT outlier

  15. notebookunverified

    Phase 0a prose refresh (full re-run, fresh captures) — NuttyCLD memory-stack article; Mizuho 800V DC confirmation; Churchill HVDC bottleneck

  16. notebookunverified

    Phase 0a prose refresh — SPCX IPO secondary effect; thesis arc validates

  17. notebookunverified

    EOW full-scan data refresh

  18. notebookunverified

    R25-C refresh: voltage-stack thesis validates, WOLF re-rate corrected

  19. notebookunverified

    Coverage gap closure: STM, MRAAY, AIXA.DE, VECO added to ai-power.json

  20. notebookunverified

    Crusoe / Lochmiller hardens incumbent-electrical bear case

  21. notebookunverified

    SPARK (NuttyCLD six-part series consolidates into perspective)

Changes to the case

2026-10-16 · EOD 2026-10-16 checkpoint: licence-dispute clause checked, not fired as written (Vicor's own patent case); dated checks corrected from receipts

Before: 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.

After: 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. · active

2026-10-16 · EOD 2026-10-16 checkpoint: licence-dispute clause checked, not fired as written (Vicor's own patent case); dated checks corrected from receipts

Before: 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.

After: 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. · active

2026-10-16 · EOD 2026-10-16 checkpoint: licence-dispute clause checked, not fired as written (Vicor's own patent case); dated checks corrected from receipts

Before: 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.

After: 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. · active

2026-10-16 · EOD 2026-10-16 checkpoint: licence-dispute clause checked, not fired as written (Vicor's own patent case); dated checks corrected from receipts

Before: 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.

After: 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. · active

2026-10-06 · EOD 2026-10-06 campaign checkpoint (book-blind Opus review)

Before: 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.

After: 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. · active

2026-10-06 · EOD 2026-10-06 campaign checkpoint (book-blind Opus review)

Before: 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.

After: 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. · active

2026-10-06 · EOD 2026-10-06 campaign checkpoint (book-blind Opus review)

Before: 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.

After: 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. · active

2026-10-06 · EOD 2026-10-06 campaign checkpoint (book-blind Opus review)

Before: 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.

After: 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. · active

2026-10-06 · EOD 2026-10-06 campaign checkpoint (book-blind Opus review)

Before: 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.

After: 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. · active

2026-09-06 · Preserve the earlier review as dated context, not a fresh classification.

Before: 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.

After: 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. · active

2026-09-06 · Attach the scope decision and evidence boundaries without promoting a working note to primary evidence.

Before: 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.

After: 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. · active

2026-09-06 · Interpret the dated primary source within this campaign’s reviewed scope.

Before: 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.

After: 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. · active

2026-09-06 · Interpret the dated primary source within this campaign’s reviewed scope.

Before: 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.

After: 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. · active

2026-09-06 · Keep distinct from generation; replace the unverified percentage-loss headline with commercial conversion tests.

Before: No prior claim

After: 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. · active