Macrontology
Research bank
Theses
Power Caps the AI Buildout — the bottleneck moved from chips to megawattsThe ABF Chokepoint Does Not Pay Its Owners — Buy the Bottleneck Fails HereASML / China-DUV displacement is overpriced — the sell-off prices near-full substitution off a 5-machine data pointThe China chokepoint moves from lithography to bonding — and the substitution trade is priced in the wrong categoryCost Per Token Is Set Outside the ChipCustom Silicon Pays a Different MerchantEMIB Shifts the Packaging Bottleneck to SubstratesHBM moat vs. DRAM commoditization — is SK Hynix's HBM-mix 'miss' actually bullish?HBM4 Qualification Concentrates Share, Not Spreads ItSilicon cannot emit light — indium phosphide is the chokepoint CPO relocates but cannot removeThe Neocloud Rent Is Consumed By The Asset Before It Reaches EquityNVIDIA's Rent Compresses Through Software, Not SiliconOptical attach is set by topology, not by shipments — linear optical TAM models are mis-specifiedThe Megawatt Is the Unit of AccountThe Upstream Is Single-Sourced and UnpricedChipflation Is a Wafer Allocation, Not a Demand ShockMemory-demand durability — does algorithmic efficiency cap the memory super-cycle?Neocloud margin re-rating — do open-source deployment + rising rental rates break the thin-margin cap?Semicap service is an installed-base annuity, and the market still prices these names as capex cyclicalsABF substrate dual-gate arbitrage: substrate converters capture the spread the market assigns upstreamAccelink: The Valuation Anomaly in China's Optical Chip IndigenizationAdvanced Packaging: Equipment Suppliers Capture the Scarcity PremiumAlphabet: The Margin Inversion Nobody PricesAmazon: The Capex-Margin InversionApplied Optoelectronics: Hyperscaler Procurement Arbitrage Driving Negative SpreadsArista: Hyperscaler capex intensity conceals margin compression riskArm: The Royalty Stream That Scales With Everyone's Margin but Its OwnCo-packaged optics transfers margin to substrate integrators, not optics specialistsConstellation: Hyperscaler Monopsony Masks Nuclear Fuel TransmissionCoWoS Constraint Prices Into the Wrong CompaniesEaton: Margin expansion telegraphs demand scarcity more than growth can showElectrical Steel Asymmetry: GE Vernova Captures Transformer Scarcity Without Steel ExposureFoxconn: Margin Squeeze Masked by AI Server Revenue MixGlobalFoundries: Capex Intensity Signals Strategic Exclusion from AI, Not WeaknessGPU collateral decay gates buildout faster than physical supplyHBM5: The Signal Integrity Tax Inverts the Memory Supplier Power BalanceTransformer lead times gate hyperscaler returns, not start datesInnolight: Negative Operating Leverage Hiding in Hyperscaler ConcentrationInterconnection queue converts hyperscaler capex into generator rentLong-context inference: memory constraint compresses cloud margin before GPU-poor plays feel itMicrosoft: Burning Cash to Rent Margin It Cannot OwnMoE inference: converted miners capture memory scarcity, hyperscalers leak itMurata: The Margin That Doesn't MoveNAND Flash: Datacenter Crowding-Out Misread as Memory Cycle RecoveryTransceiver margin expansion is a mirage: buyer concentration at 1.6T resets pricing power the market prices as durableOracle: Monopsony Rent Capture Masked by Consolidated AccountingPackage Perimeter Arbitrage: Optics Suppliers Capture Value NVIDIA Cannot RetainPalantir: The Margin-Protected Infrastructure PlayQualcomm's low capex masks an IP-only future neither consensus nor bears have pricedRack Power Density: The Liquid Cooling Adoption MirageShin-Etsu: The Hidden Rent in the DenominatorSilicon capacitor adoption arbitrages package economics invisible to component analystsSilicon Wafer Duopoly: Record Downstream Margins Signal Pricing Power Collapse, Not TightnessSK Hynix: The Customer Concentration That Validates the MoatSMIC: Bottleneck Arbitrage Disguised as National ChampionTesla: Capex Surge Hides Margin Compression at Suppliers

Silicon capacitor adoption arbitrages package economics invisible to component analysts

gen-silicon-capacitor · conviction — · status open · horizon — · as of 2026-08-10

Silicon capacitor adoption is determined by the customer's willingness to convert operating margin into package capex, not by absolute accelerator demand. Intel, AMD and NVIDIA face radically different margin structures: NVIDIA at 64% operating margin can absorb in-package solutions that Intel at -23% cannot economically justify despite identical power-delivery physics.
Rests on shares nobody discloses. 2 of 2 derived inputs move materially when the undisclosed supply weights are redrawn across their plausible range. The argument may still hold — but these figures are ranges, not points. Computed from evidence at most 21 days old (oldest input: amd).
silicon-capacitor.capex_pull>=14.6012 — 8.438 to 19.78
silicon-capacitor.demand_pull>=58.58 — 41.41 to 58.07
How to read the numbers on this page

A range instead of a point. 903 of 1,848 supply weights are not disclosed by anyone. Where redrawing them across their plausible range moves a figure by more than 25%, the figure is shown as a range and marked. A tight number is ground you can stand on; a wide one is not.

Not every premise is scored. A premise citing something the model verifies on every rebuild — a filed figure, a graph edge, a computed cell — is a PRECONDITION, not a risk. It gates the conclusion but contributes no uncertainty, because charging a conclusion for being verifiable made well-evidenced arguments look weaker than vague ones.

Composed two ways. Where several premises gate a conclusion, the figure is given as "X% if independent, Y% if they move together". They are claims about one industry, so the truth is between and nobody can say where. Treat it as an ordering device, not a calibrated probability.

Dated. Each figure is computed from facts, and the page states the age of the oldest one beneath it. The full arithmetic runs from a published model config to company revenue exposure.

Exhibits

Exhibit 1Relative performance, indexed to 100How the names in this thesis have traded against SOXX.
43387730index100 = startINTC 472SOXX 221NVDA 12012mo, indexed to 100 at start · dashed = SOXX benchmark

Series available as data/gen-silicon-capacitor.csv

The variant

Consensus

Silicon capacitors are a niche power-integrity solution for high-performance AI accelerators, beneficial where it matters but unlikely to reshape packaging economics. Demand scales with datacenter AI buildout. The technology remains incremental—a helper part—rather than a strategic shift, tracked primarily by passive-component specialists rather than packaging or semiconductor equity desks.

Variant

Silicon capacitor adoption is determined by the customer's willingness to convert operating margin into package capex, not by absolute accelerator demand. Intel, AMD and NVIDIA face radically different margin structures: NVIDIA at 64% operating margin can absorb in-package solutions that Intel at -23% cannot economically justify despite identical power-delivery physics. The technology's attach rate will bifurcate by customer profitability, not by node or workload, creating durable volume concentration at the hyperscale leaders—a margin arbitrage dynamic that component-coverage models miss entirely because they model demand as a pure function of AI unit growth.

Differentiator

Component analysts model silicon capacitors as a feature driven by technical need (power delivery, transient response). The supply chain reveals it as a *margin tax*: only customers converting rent into advanced packaging can afford in-package decoupling at scale. Customer operating margins span 87 percentage points; capex intensity follows that divergence, not the AI growth rate.

Falsifiers

Open questions

Reasoning chain

Silicon capacitor attach rates will stratify by customer operating margin rather than by accelerator performance tier, concentrating volume at NVIDIA and isolating Intel VALID
premises
  • NVIDIA Corporation — operating margin 64.0%1.00 strong
    Filed, verified.
    why

    Establishes the profit cushion available for packaging spend.

  • Intel Corporation — operating margin -23.1%1.00 strong
    Filed, verified.
    why

    Negative margin leaves no room for incremental package cost.

  • Silicon capacitor (in-package decoupling) — capex pull at least 14.6012 [8.44–19.8 depending on shares nobody discloses]1.00 strong
    Customer-weighted capex intensity already computed and bounded by supply chain.
  • Package shoreline (perimeter and cross-section contention between power, memory and data)0.75 strong
    Assumes perimeter contention remains binding enough to push decoupling on-package rather than solved by board-level design or voltage architecture changes.
  • Silicon capacitor (in-package decoupling) supplies NVIDIA Corporation1.00 strong
    Assumes NVIDIA adopts silicon capacitors in volume production; edge existence does not confirm attach rate or commercial terms.
  • Silicon capacitor (in-package decoupling) supplies Intel Corporation1.00 strong
    Intel named Powerchip exclusively for EMIB-T, but whether loss-making business sustains premium packaging components at scale is unverified.

Composed 0.75 via and over 3 gating premises · 3 supporting premises shown but not multiplied in — citing a filed figure should not cost conviction · Weakest link Package shoreline (perimeter and cross-section contention between power, memory and data) at 0.75

14.6% capex pull against 87pp margin spread implies packaging investment decisions decouple from technical roadmap alignment and follow financial capacity instead.

Powerchip's named exclusivity with Intel on EMIB-T creates single-customer concentration risk that supply-chain maps misread as a moat VALID
premises
  • Powerchip Semiconductor Manufacturing (PSMC) supplies Silicon capacitor (in-package decoupling)1.00 strong
    Recorded edge, validated.
    why

    Confirms Powerchip supplies this category.

  • Powerchip Semiconductor Manufacturing (PSMC)0.70 moderate
    Assumes exclusivity is platform-locked, not technology-locked; if Intel pivots or cuts advanced packaging spend, award evaporates.
  • Intel Corporation — operating margin -23.1%1.00 strong
    Filed.
    why

    Customer in structural loss limits ability to ramp premium packaging components.

  • Intel Corporation0.65 moderate
    Intel's packaging roadmap execution and margin recovery timeline remain uncertain; exclusivity with a restructuring customer is fragile.

Composed 0.45 via and over 2 gating premises · 2 supporting premises shown but not multiplied in — citing a filed figure should not cost conviction · Weakest link Intel Corporation at 0.65

Single-source status with a negative-margin customer inverts the usual moat logic: concentration becomes a liability if Intel cannot fund the roadmap.

Component analysts will systematically overestimate Intel's silicon capacitor TAM and underestimate NVIDIA's, misallocating supplier revenue forecasts STALE
premises
  • Silicon capacitor (in-package decoupling) — demand pull at least 58.58 [41.4–58.1 depending on shares nobody discloses]1.00 strong
    Customer-weighted growth already computed.
    why

    Establishes the blended demand signal analysts see.

  • NVIDIA Corporation — revenue growth $851.00 strong
    12x
    Filed.
    why

    NVIDIA growth is 12x Intel's, and operating margin permits packaging investment Intel cannot match.

  • Datacenter Ai0.80 strong
    Assumes datacenter AI growth continues to drive accelerator complexity; a plateau or shift to inference-optimized designs would reduce decoupling need.
  • Silicon capacitor (in-package decoupling) supplies NVIDIA Corporation1.00 strong
    Same as above: edge exists but attach-rate ramp and commercial volume unconfirmed.

Composed 0.80 via and over 1 gating premise · 3 supporting premises shown but not multiplied in — citing a filed figure should not cost conviction · Weakest link Datacenter Ai at 0.80

58.6% demand pull reflects customer revenue growth, but margin divergence means NVIDIA will convert growth into packaging adoption while Intel cannot—a split invisible in unit-based TAM models.

Sources

Write-up

Pre-filled skeleton: gen-silicon-capacitor.md