Vishay
Capacitors - Polymer Tantalum
50+
Trend upPricing upNote
AI impact on deliveries
Universe
100
Top 100 names across US and non-US passives, power, sensing, EMS, and integration.
Top US
CTS Corp
Residual alpha 100.0. The top US name is a current-sense and resistor-network bottleneck play.
Top Non-US
TT Electronics
Residual alpha 100.0. The non-US leader is a more hidden inductor and magnetics exposure.
Tightest Category
40.0w
Capacitors - Polymer Tantalum shows the highest average lead time in the screenshot-derived bottleneck set.
Constraint Vendor
Quantic Evans
Vendor constraint score 72.0. This is the hottest name in the screenshot vendor summary.
High Conviction
16
Names that still screen as the strongest alpha candidates after bottleneck, hiddenness, and crowding are combined.
Current Thesis Audit · Updated July 9, 2026
The physical bottleneck thesis is strengthening, not weakening - Murata is pushing 15-35% price hikes on AI-server MLCCs, Taiyo Yuden guides AI-server MLCC revenue +80% for FY2026 on a 1.31 book-to-bill, and >$600B of 2026 hyperscaler capex keeps demand structurally unfulfillable - but the alpha is migrating down-cap. Today's melt-up (Taiyo Yuden +20% on +389% YTD, SEMCO +18.4% on +630% YTD, Walsin and Yageo limit-up) makes the rerated MLCC tier the crowd's trade, while the US power/optics complex sold off 5-12% on Ciena sympathy and higher yields. Residual alpha now concentrates in the un-rerated power-delivery names - Sumida, CTS, TT Electronics, AcBel - and in the 800VDC/BBU architecture shift, where AcBel's immersion-cooled 800V BBU partnership is still priced at +44% YTD versus Delta's +143%. We keep the page's anti-crowding discipline: fade the limit-ups, accumulate the un-rerated PDN bottlenecks. The active ranking pass cross-checks the original thesis against current market prices, market cap, YTD rerating, AI 2027 compute/security assumptions, hyperscaler capex pressure, power constraints, HBM and advanced-packaging evidence, and physical-AI adoption signals.
Method
Residual alpha is a pricing question built on physical bottlenecks.
The ranking is designed to find names where the market has not fully capitalized the bottleneck yet. Strategic importance matters, but it matters most when it is still underappreciated.
Step 1. Measure bottleneck severity
The model starts with the screenshot set: longest lead times, repeated pricing-up arrows, and categories where shortages cluster around passives and rack power.
Step 2. Weight stack closeness
Names tied directly to current sensing, MLCCs, inductors, polymer caps, power shelves, IBCs, and rack-power hardware receive more weight than broad system beneficiaries.
Step 3. Subtract crowding
The final score penalizes crowding so the ranking prefers names where the bottleneck is real and the upside is not already fully priced.
Formula
Bottleneck severity + centrality + catalyst + AI relevance + bottleneck closeness + hiddenness/focus bonus, minus crowding and current-price rerating penalties.
bottleneck severitycentralitycatalystAI relevanceclosenesshiddennesscrowding penaltyprice rerating penalty
Bottlenecks
The data says the pain still clusters in passives and rack power.
The report is equity-facing, but the evidence base underneath it still begins with physical shortages: long lead times, categories with persistent pricing pressure, and vendors that keep showing up in constrained screenshot sets.
Highest average lead-time categories
The longest waits are still concentrated in capacitor families, current-sense, and magnetics-adjacent categories.
Names are shortened on-axis for readability. Hover for the full category name and exact lead-time values.
Most constrained screenshot vendors
This condenses the vendor-level transcriptions into a cleaner leaderboard, weighted by lead time, price pressure, and note flags.
The vendor score is a composite signal from the uploaded spreadsheet, not a market-share measure.
Capacitors - Polymer Tantalum
40.0w
Avg lead time • max 54.0w • pricing up 100%
Capacitors - Tantalum Conformals
35.5w
Avg lead time • max 40.0w • pricing up 100%
Hybrid Wet Tantalums
35.0w
Avg lead time • max 40.0w • pricing up 100%
Leaded Capacitors - Ceramic
32.2w
Avg lead time • max 40.0w • pricing up 100%
Specialty Capacitors
29.0w
Avg lead time • max 40.0w • pricing up 100%
Hybrid Aluminum Polymer Capacitors
28.5w
Avg lead time • max 33.0w • pricing up 0%
Pressure map
The exported bubble chart was too cluttered to read. This rebuild keeps the same logic but uses tooltips instead of smashed labels.
Further right means longer average lead times. Higher means more rows showing pricing pressure. Bubble size tracks the number of vendors contributing to the category.
Pressure scorecard
This replaces the unreadable heatmap with a ranked list that shows exactly why each category screens as tight.
Capacitors - Polymer Tantalum
Rank #1 on combined pressure
40.0w
Avg lead time40.0w
Pricing up100%
Trend up100%
Vendors3
Capacitors - Tantalum Conformals
Rank #2 on combined pressure
35.5w
Avg lead time35.5w
Pricing up100%
Trend up100%
Vendors2
Hybrid Wet Tantalums
Rank #3 on combined pressure
35.0w
Avg lead time35.0w
Pricing up100%
Trend up100%
Vendors1
Leaded Capacitors - Ceramic
Rank #4 on combined pressure
32.2w
Avg lead time32.2w
Pricing up100%
Trend up100%
Vendors4
Specialty Capacitors
Rank #5 on combined pressure
29.0w
Avg lead time29.0w
Pricing up100%
Trend up100%
Vendors3
High Frequency Ceramic Capacitors
Rank #6 on combined pressure
23.0w
Avg lead time23.0w
Pricing up100%
Trend up100%
Vendors1
MLCCs, Radial, Safety Caps
Rank #7 on combined pressure
23.0w
Avg lead time23.0w
Pricing up100%
Trend up100%
Vendors1
Lead time is the main driver, but categories move up the list when price hikes and persistent up-trend flags appear alongside it.
Lead-time signal board
The most telling rows from the lead-time transcription, shown as readable cards instead of a screenshot export.
Panasonic Industry
Capacitors - Polymer Tantalum
30-46
Trend upPricing upNote
AI impact on deliveries
KYOCERA AVX
Leaded Capacitors - Ceramic
40
Trend upPricing up
No additional note attached in the transcription.
KYOCERA AVX
Specialty Capacitors
40
Trend upPricing up
No additional note attached in the transcription.
KYOCERA AVX
Capacitors - Polymer Tantalum
30
Trend upPricing upNote
AI impact on deliveries; ex-Rohm devices with extended LT
KYOCERA AVX
Surface Mount General Capacitors – Ceramic *Automotive grade
28
Trend upPricing upNote
FTZ product (additional 6 weeks+ LT)
KYOCERA AVX
Capacitors - Tantalum Conformals
36
Trend upPricing up
No additional note attached in the transcription.
NIC Components
Leaded Capacitors - Ceramic
36
Trend upPricing up
No additional note attached in the transcription.
Quantic Evans
Hybrid Wet Tantalums
30-40
Trend upPricing up
No additional note attached in the transcription.
Rows bubble up here when vendor score and lead-time midpoint both matter. Additional notes are preserved directly from the pack.
Maps
The report gets stronger when you see the bottleneck as a system.
The uploaded SVGs explain the logic better than a single ranking ever could. The flow diagram maps the power path. The mindmap shows the surrounding constraint web. The layer score chart below helps connect those visuals back to the equity universe.
Supply-chain flow
Grid to rack, rack to board, board to passives, and back to system deployment.
Bottleneck mindmap
A visual map of how passives, power, cooling, packaging, and infrastructure interlock.
Layer average residual-alpha score
The layer view ties the diagrams back to the investable universe by showing where average score pools across the stack.
Named collaborationConfirmed
Eaton
Beam Rubin DSX power platform
NVIDIA Rubin DSX AI Factory
Grid-to-chip power path
Named collaborationConfirmed
Schneider Electric
AI factory power/cooling blueprint
NVIDIA Rubin AI factories
Validated with NVIDIA
Ecosystem supplyConfirmed
Vertiv
Power and cooling infrastructure
AI factories / high-density data centers
Named as NVIDIA ecosystem partner
Named product-platform fitConfirmed
Flex
800VDC Power Rack
NVIDIA Vera Rubin infrastructure
Megawatt-scale rack power
Named collaborationConfirmed
Celestica
Scale-up networking switches
AMD Helios rack-scale AI platform
Helios rack-scale AI
Named product-platform fitConfirmed
Wiwynn
Rubin NVL72 AI factory infrastructure
NVIDIA platform
Direct Rubin infrastructure demo
Named product-platform fitConfirmed
Foxconn
Full-system AI server racks
NVIDIA Vera Rubin NVL72
Modular data-center angle
Named collaborationConfirmed
LITEON
110kW power shelf
QCT Rubin NVL72 AI server solution
Shown jointly at GTC 2026
Rankings
The final leaders are not always the obvious leaders.
Switch regions to compare the two sides of the report. The bar chart makes the leaders readable, the scatter shows the crowding tradeoff, and the balance panel highlights which names still keep the most score after crowding is considered.
US leader
CTS Corp
Score 100.0 with very under-owned residual-upside labeling.
Top-10 average
64.8
Average residual alpha score across the active region's top ten names.
Very under-owned
1
Names in the active region still labeled very under-owned in the final ranking.
Direct closeness
5
Active-region names with bottleneck-closeness score of 5, which is the highest directness bucket.
Top 10 residual-alpha scores
This replaces the cramped category-axis chart with a horizontal ranking that stays legible on laptop and mobile widths.
Hover for the full company name. The active region controls both the chart and the cards below.
Residual alpha vs crowding penalty
The best names sit high on the chart without drifting too far right into obvious, rerated consensus positions.
Bubble size scales with focus bonus. Use hover for exact scores and bucket labels.
June 8 Market Refresh. Latest prices and YTD returns now come from Yahoo Finance chart data. The ranking logic stays residual-alpha first, but the visible cards and table now expose current tape pressure so users can separate thesis quality from what the market has already repriced.
Rank #1
CTS Corp
CTSPassives / sensing
100.0
residual alpha
High-conviction residual alphavery under-owned
Price
USD 59.06
YTD
+34.7%
Cap
$1.73B
Direct passive/power bottleneck supplier with lower crowding and better residual upside.
Rank #2
Benchmark Electronics
BHEEMS / industrial electronics
79.3
residual alpha
High-conviction residual alphaunderfollowed
Price
USD 81.8
YTD
+86.3%
Cap
$2.94B
Benchmark received direct HPE validation as 2026 Manufacturing Partner of the Year for scaling high-power, liquid-cooled AI server blades. That is fresh customer-level proof of the underfollowed AI infrastructure manufacturing thesis.
Rank #3
Power Integrations
POWIPower semis
67.8
residual alpha
High-conviction residual alphabalanced
Price
USD 66.46
YTD
+78.2%
Cap
$3.70B
Direct bottleneck exposure remains attractive, though less hidden than Tier A names.
Rank #4
Plexus
PLXSEMS
65.0
residual alpha
High-conviction residual alphabalanced
Price
USD 266
YTD
+74.7%
Cap
$7.11B
Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names.
Rank #5
Vishay Intertechnology
VSHPassives
64.0
residual alpha
High-conviction residual alphabalanced
Price
USD 42.19
YTD
+175.9%
Cap
$6.47B
The AI/passives backlog thesis is intact, but July brought real capital-structure overhang: the $750M equity raise and 2030 converts becoming eligible for conversion explain the -29% move and make the drawdown less clean as an alpha-restoring flush. Move down despite strong book-to-bill.
Rank #6
Littelfuse
LFUSProtection / sensing
63.3
residual alpha
Crowded / rerated benchmarkpartially rerated
Price
USD 402
YTD
+53.0%
Cap
$10.2B
Strategically important, but the stock already rerated; no longer top residual alpha.
Rank #7
Jabil
JBLPower / cooling / EMS
57.6
residual alpha
Crowded / rerated benchmarkpartially rerated
Price
USD 325
YTD
+35.3%
Cap
$34.1B
Jabil raised FY2026 revenue/EPS guidance on extremely strong AI infrastructure demand, lifted AI-related revenue to about $13.6B, and added an India AI-rack/power manufacturing platform with Adani. Crowding remains, but fundamentals outran prior expectations.
Rank #8
Regal Rexnord
RRXMagnetics / industrial
54.8
residual alpha
High-conviction residual alphapartially rerated
Price
USD 208
YTD
+42.7%
Cap
$13.9B
Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names.
Score and crowding balance
A readable cross-check on the lead cohort: strong residual alpha matters more when the crowding penalty is still modest.
CTS Corp
CTSPassives / sensing
94.0
Residual alpha100.0
Crowding penalty1.0
High-conviction residual alphavery under-owned
Benchmark Electronics
BHEEMS / industrial electronics
67.3
Residual alpha79.3
Crowding penalty2.0
High-conviction residual alphaunderfollowed
Power Integrations
POWIPower semis
49.8
Residual alpha67.8
Crowding penalty3.0
High-conviction residual alphabalanced
Plexus
PLXSEMS
47.0
Residual alpha65.0
Crowding penalty3.0
High-conviction residual alphabalanced
Vishay Intertechnology
VSHPassives
46.0
Residual alpha64.0
Crowding penalty3.0
High-conviction residual alphabalanced
Littelfuse
LFUSProtection / sensing
39.3
Residual alpha63.3
Crowding penalty4.0
Crowded / rerated benchmarkpartially rerated
The balance panel keeps the page focused on reader-facing signal and removes the last of the cluttered export graphics.
Relationships
Supply relationships make the ranking legible.
The best names in the final ranking are not random. They sit in real, named paths from grid and rack power down to server boards, current sensing, passives, and deployment partners.
Highest vendor-score rows from the lead-time transcription
| Vendor | Lead Time | Vendor Score | Signals |
|---|---|---|---|
Vishay Capacitors - Polymer Tantalum | 50+ | 100.0 | up / up |
Panasonic Industry Capacitors - Polymer Tantalum | 30-46 | 85.2 | up / up |
KYOCERA AVX Leaded Capacitors - Ceramic | 40 | 77.3 | up / up |
KYOCERA AVX Specialty Capacitors | 40 | 77.3 | up / up |
KYOCERA AVX Capacitors - Polymer Tantalum | 30 | 76.7 | up / up |
KYOCERA AVX Surface Mount General Capacitors – Ceramic *Automotive grade | 28 | 74.6 | up / up |
KYOCERA AVX Capacitors - Tantalum Conformals | 36 | 73.1 | up / up |
NIC Components Leaded Capacitors - Ceramic | 36 | 73.1 | up / up |
Quantic Evans Hybrid Wet Tantalums | 30-40 | 72.0 | up / up |
Vishay Capacitors - Tantalum Conformals | 30-40 | 72.0 | up / up |
TDK EPCOS Capacitors - Film | 24-45 | 71.5 | up / up |
KYOCERA AVX Surface Mount General Capacitors – Ceramic (Less than 1uf) | 24 | 70.4 | up / up |
KYOCERA AVX Surface Mount General Capacitors – Ceramic (greater than 1uf) | 24 | 70.4 | up / up |
WIMA Capacitors - Film | 18-24 | 67.2 | up / up |
Vishay Leaded Capacitors - Ceramic | 24-30 | 63.6 | up / up |
Vishay Specialty Capacitors | 24-30 | 63.6 | up / up |
Murata Leaded Capacitors - Ceramic | 26 | 62.5 | up / up |
Nichicon Aluminum Electrolytic | 22-30 | 62.5 | up / up |
Vishay Capacitors - Tantalum Molded | 20-30 | 61.4 | up / up |
Quantic UTC MLCCs, Radial, Safety Caps | 20-26 | 59.3 | up / up |
Quantic Eulex High Frequency Ceramic Capacitors | 20-26 | 59.3 | up / up |
TAIYO YUDEN Filters | 20-26 | 59.3 | up / up |
TAIYO YUDEN Surface Mount General Capacitors – Ceramic (greater than 1uf) | 22-24 | 59.3 | up / up |
TAIYO YUDEN Surface Mount General Capacitors – Ceramic *Automotive grade | 22-24 | 59.3 | up / up |
| Supplier | Component / Service | Recipient | Note |
|---|---|---|---|
Eaton Named collaboration | Beam Rubin DSX power platform | NVIDIA Rubin DSX AI Factory | Grid-to-chip power path |
Schneider Electric Named collaboration | AI factory power/cooling blueprint | NVIDIA Rubin AI factories | Validated with NVIDIA |
Vertiv Ecosystem supply | Power and cooling infrastructure | AI factories / high-density data centers | Named as NVIDIA ecosystem partner |
Flex Named product-platform fit | 800VDC Power Rack | NVIDIA Vera Rubin infrastructure | Megawatt-scale rack power |
Celestica Named collaboration | Scale-up networking switches | AMD Helios rack-scale AI platform | Helios rack-scale AI |
Wiwynn Named product-platform fit | Rubin NVL72 AI factory infrastructure | NVIDIA platform | Direct Rubin infrastructure demo |
Foxconn Named product-platform fit | Full-system AI server racks | NVIDIA Vera Rubin NVL72 | Modular data-center angle |
LITEON Named collaboration | 110kW power shelf | QCT Rubin NVL72 AI server solution | Shown jointly at GTC 2026 |
Delta Electronics Application-targeted | ORV3 power shelves / PSUs | AI servers | 5.5kW+ PSUs and 33kW shelves |
Bel Fuse Application-targeted | Power supplies, connectivity, magnetics | AI/cloud servers | High-density AI/cloud/edge architectures |
Vicor Application-targeted | 48V point-of-load modules | AI accelerator clusters | Current density near processors |
Monolithic Power Systems Application-targeted | High-current power modules | AI accelerator cards | 800A+ scalable current |
TDK Application-targeted | TLVR inductors | Server power supply circuits | AI-server power topology pages |
TDK Application-targeted | MLCCs | AI server power systems | AI-server power systems |
Murata Application-targeted | MLCCs, inductors, filters | AI server/data-center power networks | Power stabilization |
Samsung Electro-Mechanics Application-targeted | High-capacitance MLCCs | AI server power modules | AI-server power module stabilization |
Yageo Application-targeted | Four-terminal shunt resistors | AI server PDUs | PDU current-sense |
Yageo Application-targeted | KO-CAP, MLCC, thin-film resistors, inductors | AI/server systems | Broad AI passive portfolio |
Nippon Chemi-Con Application-targeted | Conductive polymer aluminum caps | GPU/CPU-adjacent VRM | Ultra-low ESR around processors |
Nippon Chemi-Con Application-targeted | Immersion-compatible capacitors | Liquid-cooled servers | Immersion cooling compatibility |
Panasonic Industry Application-targeted | POSCAP / polymer caps | Server and data-center power circuits | Server power circuits |
KYOCERA AVX Demand exposure | MLCCs | AI servers and GenAI systems | AI-driven demand |
Jabil Named acquisition thesis | Hanley Energy power-management stack | AI data centers | Grid-to-rack power |
Eaton / Schneider / Vertiv Ecosystem supply | Facility power and cooling | NVIDIA AI factories | Reference-design ecosystem |
NVIDIA Rubin Named downstream adopters | Rubin rack-scale systems | Microsoft / AWS / Google / OCI / CoreWeave | Cloud demand map |
NVIDIA Rubin Named downstream adopters | Rubin servers | Dell / HPE / Lenovo / Supermicro | OEM demand map |
Hyperscaler capex surge Demand signal | Multi-year AI capacity expansion | Power/cooling/passive supply chain | Capex quadrupled since GPT-4 |
Alphabet / Amazon / Microsoft capex surge Demand signal | AI demand exceeded available supply | Upstream suppliers | Supply exceeded demand |
AI data-center delays Macro bottleneck | Utility, labor and equipment bottlenecks | Power/cooling/grid suppliers | ~40% sites may face delays |
AI demand in China Macro bottleneck | Equipment, passive components, talent | Passive suppliers | Passives explicitly referenced |
Schneider / Rubin cooling roadmap Density signal | 150kW+ racks and liquid cooling | Cooling stack suppliers | Liquid cooling effectively required |
RBC data-center power struggle Macro bottleneck | Power availability as bottleneck | Grid/power suppliers | Power first, not just chips |
ITIF AI data-center power concerns Macro bottleneck | Reliability and power conditioning | UPS / capacitor / power-quality chain | System-level reliability |
Switchgear / transformers / busway Layered inference | Rack power feeds | Power shelves / PSUs | Grid to rack |
Power shelves / PSUs Layered inference | 48V buses and VRMs | Motherboards / accelerator cards | Rack to board |
VRMs and power planes Layered inference | MLCCs, polymer caps, inductors, current sense | GPU / CPU packages | Board-level component stack |
Substrates / PCBs / backplanes Layered inference | Accelerators, NICs and switches | Rack integrators / ODMs | Board to rack |
ODMs / EMS Layered inference | AI racks, storage and network fabrics | Hyperscalers / AI labs | Rack to deployment |
Explorer
Search the full ranking.
The native explorer is built on the final top-100 master file. Filter by region, bucket, or layer, then sort by rank, score, positive rank change, or lower crowding penalty.
49 rows
| Rank | Company | Bucket | Δ Rank | Price | YTD | Score | Cases | Crowding | Residual Upside | Ranking Note |
|---|---|---|---|---|---|---|---|---|---|---|
| #3 | CTS Corp CTS • Passives / sensing | High-conviction residual alpha | -2 | USD 59.06 | +34.7% | 100.0 | 1.0 | very under-owned | Direct passive/power bottleneck supplier with lower crowding and better residual upside. | |
| #7 | Benchmark Electronics BHE • EMS / industrial electronics | High-conviction residual alpha | -5 | USD 81.8 | +86.3% | 79.3 | 2.0 | underfollowed | Benchmark received direct HPE validation as 2026 Manufacturing Partner of the Year for scaling high-power, liquid-cooled AI server blades. That is fresh customer-level proof of the underfollowed AI infrastructure manufacturing thesis. | |
| #12 | Power Integrations POWI • Power semis | High-conviction residual alpha | -9 | USD 66.46 | +78.2% | 67.8 | 3.0 | balanced | Direct bottleneck exposure remains attractive, though less hidden than Tier A names. | |
| #14 | Plexus PLXS • EMS | High-conviction residual alpha | -10 | USD 266 | +74.7% | 65.0 | 3.0 | balanced | Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names. | |
| #15 | Vishay Intertechnology VSH • Passives | High-conviction residual alpha | -8 | USD 42.19 | +175.9% | 64.0 | 3.0 | balanced | The AI/passives backlog thesis is intact, but July brought real capital-structure overhang: the $750M equity raise and 2030 converts becoming eligible for conversion explain the -29% move and make the drawdown less clean as an alpha-restoring flush. Move down despite strong book-to-bill. | |
| #16 | Littelfuse LFUS • Protection / sensing | Crowded / rerated benchmark | -11 | USD 402 | +53.0% | 63.3 | 4.0 | partially rerated | Strategically important, but the stock already rerated; no longer top residual alpha. | |
| #22 | Jabil JBL • Power / cooling / EMS | Crowded / rerated benchmark | -16 | USD 325 | +35.3% | 57.6 | 4.0 | partially rerated | Jabil raised FY2026 revenue/EPS guidance on extremely strong AI infrastructure demand, lifted AI-related revenue to about $13.6B, and added an India AI-rack/power manufacturing platform with Adani. Crowding remains, but fundamentals outran prior expectations. | |
| #23 | Regal Rexnord RRX • Magnetics / industrial | High-conviction residual alpha | -15 | USD 208 | +42.7% | 54.8 | 4.0 | partially rerated | Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names. | |
| #30 | Rogers Corp ROG • Materials | Secondary residual alpha | -18 | USD 135 | +47.2% | 48.4 | 4.0 | partially rerated | Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names. | |
| #31 | Bel Fuse BELFB • Power delivery / connectivity | Crowded / rerated benchmark | -20 | USD 259 | +49.7% | 48.0 | 5.0 | already rerated | Strategically important, but the stock already rerated; no longer top residual alpha. | |
| #32 | Sanmina SANM • EMS / integration | Crowded / rerated benchmark | -23 | USD 210 | +31.6% | 47.8 | 4.0 | partially rerated | A -5.26% one-day sell-off driven purely by the 30-year yield breaching 5% — not by any crack in the AI/cloud integration thesis — trims the rerating premium this alpha page penalizes and modestly improves entry asymmetry versus yesterday's hold at a higher price. | |
| #34 | Microchip Technology MCHP • Power / embedded | Secondary residual alpha | -24 | USD 85.49 | +31.5% | 46.4 | 4.0 | partially rerated | Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names. | |
| #36 | ON Semiconductor ON • Power semis | Secondary residual alpha | -19 | USD 93.79 | +65.4% | 45.6 | 4.0 | partially rerated | On a residual-alpha page, a -6.3% flow-driven flush paired with genuinely new power-bottleneck evidence (NVIDIA 800VDC collaboration, GaNEXUS launch) improves asymmetry versus rank-37 peers that only de-rated. | |
| #37 | Powell Industries POWL • Switchgear | Crowded / rerated benchmark | -16 | USD 232 | +97.3% | 45.5 | 4.0 | partially rerated | Strategically important, but the stock already rerated; no longer top residual alpha. | |
| #43 | Analog Devices ADI • Analog / monitoring | Secondary residual alpha | -30 | USD 385 | +40.8% | 40.7 | 4.0 | partially rerated | Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names. | |
| #45 | Monolithic Power Systems MPWR • Power semis | Crowded / rerated benchmark | -30 | USD 1,316 | +40.5% | 40.5 | 5.0 | already rerated | On a residual-alpha list, a fresh fiduciary-duty investigation plus ~$380M of insider selling at peak valuation is new negative evidence layered on an already 'crowded/rerated benchmark' name; today's -4.6% is governance-driven, not a thesis-intact flush that improves asymmetry. | |
| #46 | Navitas Semiconductor NVTS • Power semis | Secondary residual alpha | -19 | USD 13.36 | +59.4% | 39.8 | 4.0 | partially rerated | Navitas still has valuable 800V/GaN/SiC AI-power optionality, but Wolfspeed sued over patents covering broad GaNFast/GaNSlim/GaNSafe/GeneSiC product lines. Legal/injunction/licensing risk is now thesis-relevant, not merely a no-news de-crowding flush. | |
| #48 | Modine MOD • Thermal management | Crowded / rerated benchmark | -29 | USD 236 | +67.3% | 39.5 | 4.0 | partially rerated | Strategically important, but the stock already rerated; no longer top residual alpha. | |
| #49 | Vishay Precision Group VPG • Current sensing | Crowded / rerated benchmark | -33 | USD 108 | +174.8% | 39.2 | 4.0 | partially rerated | Strategically important, but the stock already rerated; no longer top residual alpha. | |
| #50 | Hubbell HUBB • Power infrastructure | Secondary residual alpha | -36 | USD 481 | +3.8% | 37.9 | 4.0 | partially rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #51 | Texas Instruments TXN • Analog / power | Secondary residual alpha | -33 | USD 301 | +69.7% | 36.0 | 4.0 | partially rerated | Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names. | |
| #52 | Lumentum LITE • Optics / photonics | Monitor / benchmark | -27 | USD 707 | +83.1% | 35.2 | 4.0 | partially rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #53 | Vicor VICR • Power delivery | Crowded / rerated benchmark | -24 | USD 263 | +125.2% | 34.8 | 5.0 | already rerated | Strategically important, but the stock already rerated; no longer top residual alpha. | |
| #54 | Rambus RMBS • Memory interfaces | Monitor / benchmark | -34 | USD 110 | +10.4% | 34.3 | 4.0 | partially rerated | Residual-alpha-vs-passives lens: a -12% flow-driven flush with the HBM4E/memory-interface thesis intact further compresses YTD and unwinds the crowding that anchored prior holds. No RMBS-specific fundamental damage, so the de-rate is alpha-restorative, not thesis erosion. Extends prior 'up'. | |
| #55 | MACOM Technology Solutions MTSI • Networking analog / optical | Monitor / benchmark | -27 | USD 305 | +74.5% | 33.9 | 4.0 | partially rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #56 | Advanced Energy Industries AEIS • Power conversion | Monitor / benchmark | -33 | USD 294 | +32.3% | 33.0 | 5.0 | already rerated | Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names. | |
| #58 | Carrier Global CARR • Cooling / HVAC | Secondary residual alpha | -36 | USD 66.74 | +24.7% | 32.1 | 4.0 | partially rerated | On a crowding-sensitive residual-alpha list, a +4.7% pop on conference talk with zero new fundamental evidence, stacked on +31.8% YTD, consumes residual alpha rather than creating it; the name is already flagged as farther from the passive/power bottleneck stack. | |
| #60 | Cummins CMI • Power generation | Monitor / benchmark | -36 | USD 669 | +28.2% | 30.7 | 4.0 | partially rerated | Residual-alpha vs passives: a concrete 2 GW AI data-center gas-gen order (Circe), raised 2026 and 2030 guidance, and a dedicated 4 MW data-center engine pull Cummins meaningfully closer to the power-bottleneck core this page rewards - more than 'Monitor/benchmark' implies. Genuine fundamental step-u | |
| #63 | Pure Storage PSTG • Storage | Monitor / benchmark | -37 | USD 67.8 | -1.8% | 29.2 | 4.0 | partially rerated | Hyperscaler DirectFlash plan beaten, FY26 guide raised and FY27 hyperscaler investment signaled - the AI storage/data-path thesis is now validated, not speculative. With negative YTD, asymmetry on this crowding-sensitive list improves. Still farther from the passive/power core the page favors. | |
| #64 | TTM Technologies TTMI • PCB / backplane | Crowded / rerated benchmark | -30 | USD 145 | +105.8% | 29.2 | 5.0 | already rerated | Strategically important, but the stock already rerated; no longer top residual alpha. | |
| #68 | Corning GLW • Fiber / materials | Secondary residual alpha | -38 | USD 184 | +103.0% | 26.6 | 4.0 | partially rerated | Genuinely new evidence since Jun 6 (third hyperscaler megadeal, Amazon) plus a -9.35% flow-driven flush with the thesis intact improves residual asymmetry at an already crowding-discounted rank 49; the ~17% give-back from recent highs de-crowds the trade somewhat. | |
| #69 | Generac GNRC • Backup power | Monitor / benchmark | -36 | USD 237 | +67.7% | 25.0 | 4.0 | partially rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #70 | Caterpillar CAT • Power generation | Monitor / benchmark | -39 | USD 948 | +58.4% | 23.0 | 4.0 | partially rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #71 | Ciena CIEN • Transport networking | Monitor / benchmark | -39 | USD 441 | +79.2% | 22.4 | 4.0 | partially rerated | On a crowding-sensitive alpha list, a flow-driven -8.3% flush on top of a ~31% post-earnings derating - while the fundamentals actually improved (record $7.7B backlog, raised FY guide) - restores some asymmetry that yesterday's hold lacked. Move is modest because Ciena remains peripheral to the pass | |
| #72 | Vertiv VRT • Facility power / cooling | Crowded / rerated benchmark | -37 | USD 318 | +81.0% | 19.2 | 5.0 | already rerated | On this crowding-sensitive alpha list, a -5.3% day inside a ~12% weekly washout with the bottleneck thesis fully intact (analysts raising targets into the drop, FY26 guidance unchanged) modestly improves residual asymmetry by flushing momentum money. It does not de-crowd a $115B name up 62% YTD, so | |
| #73 | Hewlett Packard Enterprise HPE • Servers / networking | Monitor / benchmark | -37 | USD 44.67 | +84.8% | 19.2 | 4.0 | partially rerated | Two-day, ~13% flow-driven dump (-27% from the June 2 high) with the fundamental thesis actually strengthened by the FQ2 guidance raise improves residual asymmetry versus yesterday's hold, which predated today's leg down. Still only a modest up: on a passives-alpha lens HPE remains a fully discovered | |
| #79 | Super Micro Computer SMCI • Servers / racks | Monitor / benchmark | -40 | USD 28.17 | -9.0% | 11.7 | 5.0 | already rerated | On a passives/power-bottleneck lens SMCI was already peripheral, and the $7B dilutive raise plus cancellable $39B order book and Goldman's 27%-customer concentration warning directly attack the 'tight-supply, central-node' claim. Residual alpha vs passives names worsens. | |
| #80 | Broadcom AVGO • Networking / custom silicon | Monitor / benchmark | -43 | USD 389 | +11.8% | 11.7 | 5.0 | already rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #81 | Amphenol APH • Connectivity | Crowded / rerated benchmark | -43 | USD 158 | +13.3% | 11.4 | 5.0 | already rerated | On a crowding-sensitive alpha list, an ~8.7% two-day rally with zero new fundamental evidence is pure post-rerating chase that further erodes residual alpha; the name is already bucketed 'Crowded / rerated benchmark' at a $182B cap, so it should slip below smaller, undiscovered bottleneck names. | |
| #82 | Arista Networks ANET • Networking | Monitor / benchmark | -42 | USD 181 | +35.5% | 9.9 | 5.0 | already rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #83 | Micron Technology MU • Memory | Monitor / benchmark | -42 | USD 949 | +200.8% | 7.7 | 4.0 | partially rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #86 | EMCOR Group EME • Electrical / mechanical construction | Monitor / benchmark | -44 | USD 769 | +20.4% | 5.0 | 5.0 | already rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #89 | Quanta Services PWR • Electrical construction | Monitor / benchmark | -45 | USD 666 | +51.5% | 4.0 | 5.0 | already rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #90 | Astera Labs ALAB • AI connectivity | Monitor / benchmark | -47 | USD 393 | +119.0% | 3.7 | 5.0 | already rerated | Balanced AI supply-chain exposure, but not as mispriced as the top residual-alpha names. | |
| #92 | Comfort Systems USA FIX • Mechanical / HVAC | Monitor / benchmark | -46 | USD 1,685 | +67.9% | 2.3 | 5.0 | already rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #94 | Dell Technologies DELL • Servers | Monitor / benchmark | -49 | USD 432 | +238.0% | 0.2 | 4.0 | partially rerated | Yesterday's downgrade was purely rerating discipline; today's -7.5% flow-driven washout (sector de-grossing plus Silver Lake supply overhang, ~20% off the post-earnings high in a week) with the $51.3B AI backlog thesis fully intact partially restores asymmetry. Only a modest move up: at $241B cap an | |
| #95 | Bloom Energy BE • Onsite power | Crowded / rerated benchmark | -47 | USD 254 | +157.7% | 0.0 | 5.0 | already rerated | Strategically important, but the stock already rerated; no longer top residual alpha. | |
| #96 | Marvell Technology MRVL • Networking / custom silicon | Monitor / benchmark | -49 | USD 232 | +159.2% | 0.0 | 5.0 | already rerated | Useful AI exposure, but farther from the passive/power bottleneck stack. | |
| #97 | Western Digital WDC • Storage | Monitor / benchmark | -48 | USD 550 | +193.2% | 0.0 | 4.0 | partially rerated | Residual-alpha lens: sold-out 2026 HDD capacity, multi-year LTAs to 2028 and record margins are a real fundamental step-change, not just tape. Even as a 'monitor/benchmark' name farther from the passive/power core, the supply-tightness signal merits an up nudge on news, separate from the crowding/YT |
Sources & Downloads
Everything needed to audit the final ranking is available from the page.
The page surfaces the source trail behind the ranking and gives interested readers a lightweight way to inspect the raw data without turning the blog into a download center.
Reader notes
Most readers only need the narrative, charts, and source table below. For anyone who wants the raw numbers, the two links here cover the full ranked universe and the exact source index behind the page.
| Type | Title | Used For |
|---|---|---|
| Official | NVIDIA DSX AI factory reference design | Names Eaton, Schneider Electric, Vertiv and energy leaders in DSX ecosystem. |
| Official | TDK MLCC solutions for AI server power systems | Explains MLCCs as key design constraints in AI-server PSUs and IBCs. |
| Official | Nippon Chemi-Con generative AI and aluminum electrolytic capacitors | Maps the power path from AC/DC and UPS down to server and GPU-adjacent power. |
| Official | Murata data center server page | Shows polymer aluminum electrolytic capacitors and DC-DC converters in server power. |
| Official | Murata OCP / data center power page | States OCP-compatible PSUs and power shelves for AI servers and high-density data centers. |
| Official | LITEON + QCT GTC 2026 rack solution | 110kW power shelf with QCT Rubin NVL72 solution. |
| Official | LITEON NVIDIA GTC 2026 data center portfolio | 800VDC power rack, 110kW power shelf, 2.1MW in-row CDU. |
| Official | Delta ORV3 33kW power system | AI-server power shelf and 5.5kW PSU stack. |
| Official | Delta APEC 2024 AI server power announcement | 33kW / 18kW shelves and vertical power delivery at GPU/CPU/ASIC level. |
| Official | Yageo AI page | Yageo positions its products as supporting AI power needs. |
| Official | Yageo AI-server PDU shunt resistor release | Four-terminal shunt resistors for AI server PDUs. |
| Official | CTS current sensors | CTS cites data centers and power distribution among current sensor applications. |
| Official | CTS resistor networks | Resistor arrays for FPGA, PCIe, backplane and high-density interfaces. |
| Official | CTS shunts | Precision shunts for AC/DC converters and battery/current monitoring. |
| Official | Sumida corporate page | Power inductors, transformers, EMC coils and magnetic materials. |
| Official | Sumida power inductors | Power inductor catalog and applications. |
| Official | Walsin about | One-stop passive component supplier with data-center exposure. |
| Official | Walsin capacitors | MLCC portfolio including high-capacitance and high-voltage families. |
| Official | Vishay tantalum portfolio | Broad tantalum offering across molded, conformal, low-ESR and infrastructure uses. |
| Official | Vishay polymer tantalum release | Polymer tantalum with low ESR for smoothing and filtering. |
| Official | Belden data centers | Data-center cabling, power distribution and monitoring context. |
| Official | Plexus manufacturing overview | Complex product manufacturing and supply-chain services. |
| Official | Kimball industrial electronics | Durable industrial electronics manufacturing and controls. |
| Market data | Bel Fuse statistics | Used to penalize already-rerated winners. |
| Market data | Delta Electronics statistics | Used to penalize already-rerated winners. |
| Market data | Bel Fuse market-cap history | 1-year market-cap change. |
| Market data | Delta Electronics market-cap history | 1-year market-cap change. |
| Market data | CTS statistics | Residual alpha calibration for a direct passive/sensing name. |
| Market data | Benchmark statistics | Residual alpha calibration for an underfollowed EMS name. |
| Market data | Walsin statistics | Residual alpha calibration for an underfollowed passive supplier. |
| Market data | Sumida statistics | Residual alpha calibration for a smaller inductor supplier. |
| Market data | Vishay overview | Residual alpha calibration for a diversified passive supplier. |
| Market data | Nippon Chemi-Con overview | Residual alpha calibration for capacitor supplier. |
| Market data | Nichicon overview | Residual alpha calibration for capacitor supplier. |
| Market data | TDK overview | Residual alpha calibration for a direct passive supplier. |
| Market data | Panasonic overview | Residual alpha calibration for polymer/tantalum/passive exposure. |
| Market data | discoverIE overview | Residual alpha calibration for underfollowed power/connectivity group. |
| Market data | Yahoo Finance chart endpoint | July 9, 2026 latest prices and YTD return refresh for public tickers. |
This report is a research tool, not financial advice. Residual alpha is an inference about mispricing and crowding, not a guarantee about future returns.