SemiconductorX > Spotlights
Semiconductor Spotlights
The Spotlights are SemiconductorX's company and program-level supply chain deep-dives. Where the Sectors pillar maps semiconductor demand by end market and the Chip Types pillar maps supply by device category, Spotlights follow the specific companies and programs whose supply chain decisions, market concentration, or geopolitical position make them disproportionately important to understanding the silicon economy. A handful of companies — TSMC, NVIDIA, ASML, Wolfspeed — exercise supply chain leverage far beyond their revenue rank because they occupy sole-source or near-sole-source positions in critical supply chains the entire industry depends on. Others — Tesla Terafab, Huawei HiSilicon, SpaceX — are building vertically integrated silicon programs whose outcomes will define competitive landscapes through 2030 and beyond.
Spotlights organize into four categories: Foundry & Equipment (the manufacturing supply chain); AI & Compute (the dominant demand-side companies); Automotive, Robotics & Industrial (electrification and physical AI silicon programs); and Sector Programs (mission-specific supply chain case studies). Each spotlight covers supply chain position, key products, foundry dependencies, geographic concentration, geopolitical exposure, supply chain bottlenecks, and outlook. The navigation logic: Sectors identify the constraint, Chip Types map the device supply chain, Spotlights profile the companies whose decisions determine whether and when the constraint resolves.
Foundry & Equipment
The foundry and equipment tier sits at the top of the semiconductor supply chain hierarchy. These companies do not design the chips the world uses — they manufacture them, or supply the equipment required to manufacture them. Their leverage derives from concentration: TSMC manufactures approximately 90% of chips at sub-5nm, ASML supplies 100% of EUV lithography scanners, and no credible alternative to either exists within this decade. Supply chain analysis at any other tier is incomplete without understanding the constraints these companies impose on the system.
| Spotlight | HQ | Why it matters to SX | Key coverage |
|---|---|---|---|
| TSMC | Hsinchu, Taiwan | The single most critical node in the global semiconductor supply chain. ~90% sub-5nm market share. NVIDIA overtook Apple as TSMC's #1 customer in 2025 at ~19% revenue. CoWoS capacity (120-130K wafers/month target end 2026) is the binding AI GPU packaging constraint. N2 volume production underway; both Taiwan N2 fabs sold out all 2026. Arizona $165B buildout accelerating. | N2/N3/A16 node roadmap; CoWoS packaging monopoly and expansion; NVIDIA 60% CoWoS allocation; Arizona Fab 21 complex (Fab 1 in production, Fab 2 N3 accelerated to 2027); Japan JASM; Germany ESMC; Apple/NVIDIA/AMD/Broadcom customer concentration; Taiwan geopolitical risk; $52-56B 2026 CapEx |
| Samsung Semiconductor | Suwon / Pyeongtaek, South Korea | The most vertically integrated semiconductor company in the world — Memory, Foundry, and LSI under one structure. HBM4 comeback: first to ship (February 2026), 4nm base die + 1c DRAM technology leadership, ~30% NVIDIA Vera Rubin allocation, pricing parity restored with SK Hynix. SF2 (2nm GAA) foundry recovery. Critical internal conflict: >50% of Pyeongtaek foundry capacity redirected to HBM4 base die over external customers. Taylor Texas pilot ops February 2026. | HBM4 (4nm base die, 1c DRAM, ~30% Vera Rubin allocation, first shipments Feb 2026); HBM4E (4.0 TB/s, 16 Gbps, GTC 2026); SF2/SF2P 2nm GAA yield trajectory (50-60%, targeting 60-70%); Taylor Texas Fab 1 ($25B, pilot ops Feb 2026); Tesla AI6 $16.5B anchor; Groq 3 LPU on Samsung 4nm; Pyeongtaek P4 1c DRAM conversion; OpenAI Titan chip HBM4 exclusive |
| Intel Foundry | Santa Clara, California, US | The only US-headquartered company manufacturing sub-5nm chips. Intel 18A (GAA RibbonFET + PowerVia backside power) in HVM at Fab 52 Arizona since January 2026. Panther Lake launched CES 2026 (180 TOPS NPU). Terafab partnership announced April 7, 2026 — most commercially significant IFS win since IDM 2.0. NVIDIA invested $5B in Intel. Only credible near-term US-domiciled alternative to TSMC for advanced node fabless customers. | Intel 18A (RibbonFET + PowerVia, HVM Jan 2026); Panther Lake / Clearwater Forest; 18A yield (~60%, profitable threshold ~70-80%); Terafab partnership (April 7 2026, IFS anchor arrangement); IFS customers (Microsoft, Amazon, Apple discussed); Ohio fab delay (2030+); 14A roadmap (High-NA EUV, 2027); NVIDIA $5B stake; AI inference pivot (Crescent Island) |
| Wolfspeed | Durham, North Carolina, US | Emerged from Chapter 11 September 29, 2025 (prepackaged, 91 days; debt cut 70% to $4.6B). Durham 150mm device fab closed November 2025; all production consolidated at Mohawk Valley 200mm. World's #1 SiC substrate supplier (33.7% market share). AI datacenter SiC revenue +50% sequentially Q2 FY2026. Nine-market demand convergence creates structural demand that no bankruptcy removes. | Chapter 11 emergence (Sept 2025) — causes, terms, customer impact; Mohawk Valley 200mm consolidation; Durham closure; nine-market SiC demand mapping; Chinese SiC competition (SICC 17.1%, TanKeBlue 17.3%); AI datacenter SiC diversification; 300mm SiC demonstration; CHIPS Act 48D credit ($698.6M received Dec 2025) |
| ASML (pipeline) | Veldhoven, Netherlands | Sole supplier of EUV lithography scanners globally — the most concentrated equipment chokepoint in the supply chain. ~50-60 EUV systems/year production rate is the physical ceiling on global leading-edge fab capacity addition. High-NA EUV (~20 systems/year at $380M each) defines sub-2nm. Carl Zeiss SMT optics is the binding sub-supply constraint within ASML's own chain. | EUV scanner technology and annual production rate; High-NA EUV EXE:5000 ramp; Carl Zeiss SMT optics dependency; EUV export restrictions (China ban, Netherlands/US coordination); DUV NXT series; TSMC/Samsung/Intel EUV allocation; capacity math vs global fab expansion plans |
| GlobalFoundries (pipeline) | Santa Clara, California, US | World's largest specialty foundry by revenue after exiting the leading-edge race in 2018. GF Fab 9 (Malta NY) SiGe BiCMOS is the primary Western source for mmWave radar and defense RF ICs. FD-SOI 22FDX at Dresden for ultra-low-power IoT. GF is the supply chain answer to "who makes the chips TSMC won't" — specialty processes with no commercial alternative. | SiGe BiCMOS Fab 9 (Malta NY — defense radar, 5G mmWave, 6G research); FD-SOI 22FDX (Dresden — IoT, ultra-low-power); 12LP+ FinFET; RF-CMOS; CHIPS Act funding; GF Singapore; defense and government supply chain qualification |
AI & Compute
The AI and compute tier covers the companies whose silicon decisions are driving the largest demand signals in the semiconductor supply chain today. NVIDIA's GPU concentration mirrors TSMC's foundry concentration on the demand side: ~80% of AI training accelerator revenue concentrated in one company, manufactured at one foundry, packaged with one advanced packaging process, using one primary HBM supplier. The hyperscaler custom ASIC programs and AMD's MI400 series represent the structural response — but they compete for the same upstream TSMC, CoWoS, and HBM supply chains rather than relieving them.
| Spotlight | HQ | Why it matters to SX | Key coverage |
|---|---|---|---|
| NVIDIA | Santa Clara, California, US | ~80% AI training accelerator market share; $1 trillion Blackwell + Vera Rubin order backlog through 2027. TSMC's #1 customer at ~19% revenue; holds ~60% of TSMC CoWoS allocation. Vera Rubin GPU (TSMC N3, 8 HBM4 stacks, 576GB Superchip); production target revised 2M → 1.5M units due to HBM4 delays; ODM mass production shifted June → September. HBM4: SK Hynix 70%, Samsung 30%, Micron HVM confirmed GTC March 2026. CUDA ecosystem is the durable moat — not the GPU itself. | Blackwell / Vera Rubin / Rubin Ultra / Feynman roadmap; HBM4 three-supplier qualification race; CoWoS-L packaging proprietary bottleneck; Groq 3 LPU ($20B acqui-hire, Samsung 4nm, 1-2K racks 2026); Space-1 Vera Rubin orbital compute (GTC 2026); China market loss (H20 ban April 2025, ~zero share); CUDA ecosystem moat; NVLink Fusion; NVIDIA $5B Intel stake; $2B Marvell investment |
| AMD | Santa Clara, California, US | The credible NVIDIA alternative and most successful server CPU turnaround in semiconductor history. MI350X (CDNA 4, TSMC N3P, 288GB HBM3e) in production. MI400/Helios rack (TSMC N2, 432GB HBM4, 72 MI455X, 3 exaFLOPS, Q3 2026) is AMD's most competitive AI platform yet. OpenAI 6GW + Meta 6GW custom MI450 are the largest non-NVIDIA AI chip commitments ever made. EPYC Turin ~40% x86 server share. Venice (256-core Zen 6, TSMC N2) H2 2026. 100% TSMC dependent. | MI350X (N3P, HBM3e, in production); MI455X / Helios rack (N2, HBM4, UALink, Q3 2026); EPYC Venice (256-core Zen 6, N2); OpenAI 6GW + equity option; Meta 6GW custom MI450 (19.6 TB/s HBM4); Oracle lead partner; ROCm 7.2 ecosystem; chiplet architecture (XCD + IOD); Strix Halo; Radeon RX 9000; TSMC N2 allocation competition; AI accelerator share trajectory (4-9% → 18% projected) |
| Huawei / HiSilicon | Shenzhen, China | The defining case study in export control policy and domestic semiconductor self-sufficiency. Ascend 910C (SMIC N+2, ~60% H100 inference) shipped 805K units in 2025. CloudMatrix 384 surpasses NVL72 aggregate PFLOPs at 4.1x the power. Ascend 950PR (1.56 PFLOPS FP4, 750K units planned 2026, ByteDance $5.6B). HBM is the binding constraint — not SMIC wafer capacity. TSMC die bank (2.9M dies via Sophgo) exhausted. H20 ban April 2025 leaves Ascend as the only legal AI accelerator for Chinese buyers. | Kirin 9000S/9020 (SMIC N+2 DUV-only 7nm — proof of concept for domestic manufacturing); Ascend 910C/920/950PR roadmap; CloudMatrix 384 architecture; HBM binding constraint (CXMT ~2M stacks 2026 vs >1M die capacity); TSMC die bank exhaustion and SMIC transition; SMIC process ceiling (DUV-only ~7nm equivalent); CANN vs CUDA; Beijing GPU directive (Sept 2025) |
| ASML | Veldhoven, the Netherlands | Most cross-referenced missing spotlight across mfg-process-nodes.html, fab clusters, and all leading-edge foundry spotlights. The EUV scanner production rate is the binding physical ceiling on all leading-edge capacity addition — no other single fact in SX is referenced as frequently without a home page. | EUV scanner technology (CO2 laser + tin droplet plasma); High-NA EUV EXE:5000 production ramp (~20/year, $380M each); Carl Zeiss SMT optics as sub-supply constraint; export control mechanics (EUV to China — Netherlands/US coordination); DUV NXT series; TSMC/Samsung/Intel as primary EUV customers; annual production rate math vs global fab expansion demand |
| Apple Silicon (pipeline) | Cupertino, California, US | TSMC's second-largest customer (~18-20% revenue); anchor N2 customer (pre-purchased >50% initial supply). C1 modem (iPhone 16e) displaces 20-25% of Qualcomm revenue — most commercially significant supply chain transition in smartphone history. Apple Intelligence NPU drives N2 demand independently of raw performance competition. Four 2nm chipsets in development for iPhone 17/A19 cycle. | A18/A18 Pro (N3E); M4 family (N3E); C1 modem (first in-house 5G modem); Apple Intelligence NPU; N2 anchor customer economics (>50% initial supply); TSMC PDK co-development depth; Sony CMOS image sensor dependency; Qualcomm displacement timeline; Vision Pro R1 |
| Broadcom (pipeline) | San Jose, California, US | Dominant Ethernet switch ASIC (Tomahawk/Jericho); largest custom AI ASIC co-design partner for hyperscalers (Google TPU, Meta MTIA, OpenAI Titan); holds ~15% of TSMC CoWoS allocation, rising rapidly toward top-3 customer. NVLink Fusion partnership with NVIDIA. Custom ASIC programs are a structural TSMC/CoWoS demand addition not a NVIDIA substitution. | Tomahawk 5 / Jericho3-AI Ethernet switch ASICs; custom ASIC co-design (Google TPU v5/v6, Meta MTIA, OpenAI Titan); NVLink Fusion (NVIDIA partner); Spectrum-X CPO Ethernet; Ultra Ethernet Consortium; rising TSMC CoWoS allocation share; VMware integration; Google TPU 2031 long-term agreement |
| Qualcomm (pipeline) | San Diego, California, US | Dominant 5G modem IP licensor and Android flagship SoC supplier facing the most commercially significant customer disruption in its history — Apple C1 modem displacing 20-25% of handset revenue. Snapdragon X Elite and Snapdragon Ride are the diversification bets. Samsung 2nm discussions ongoing. IP licensing model is structurally durable regardless of hardware share shifts. | Snapdragon 8 Elite (TSMC N3E); Snapdragon X Elite/Plus (Windows ARM PC); Apple C1 modem displacement — revenue impact and timeline; X80 5G modem roadmap; QTM mmWave AiP; Snapdragon Ride Elite ADAS; $45B+ automotive pipeline; IP licensing revenue durability; Samsung 2nm discussions |
Automotive, Robotics & Industrial
The automotive and industrial tier covers the companies whose silicon programs are driving the most structurally complex supply chain dynamics in the physical economy. Tesla's Terafab vertical integration program, the humanoid robot analog/mixed-signal supply gap, Infineon's SiC dominance in automotive power, and the space compute semiconductor stack are all generating demand signals the semiconductor industry was not designed to accommodate at current scale or on current timelines.
| Spotlight | HQ | Why it matters to SX | Key coverage |
|---|---|---|---|
| Tesla EV & AI | Austin, Texas, US | The most vertically integrated automotive semiconductor program. Three-chip family: AI5 (TSMC Arizona + Samsung Taylor dual-source), AI6 (Samsung Taylor captive, $16.5B through 2033), AI7/D3 (rad-tolerant, Terafab orbital target). Terafab (Tesla/SpaceX/xAI JV, announced March 21 2026, Intel joining April 7): the most discussed new fab of 2026. Optimus robot drives humanoid semiconductor demand independently of automotive programs. | FSD AI5 (TSMC Arizona + Samsung Taylor); AI6 (Samsung Taylor SF2P, dedicated captive, $16.5B through 2033); AI7/D3 (orbital rad-tolerant, Terafab); Terafab structure (Intel IFS anchor, $20-25B, North Campus Giga Texas); CyberCab robotaxi silicon; Optimus robot semiconductor stack; Samsung Taylor captive fab mechanics; Tesla acquisition of Intel speculation (CFIUS analysis) |
| Humanoid Robots | Multiple (Tesla Austin TX; Figure Sunnyvale CA; Agility Robotics Corvallis OR; 1X Moss Norway; Unitree Hangzhou China) | The most analytically undercovered semiconductor demand story of the 2026-2030 period. 300-400 analog and mixed-signal ICs per robot — GaN motor drive ICs, magnetic position encoders, MEMS IMUs, force-torque sensors, BMS ICs, current sense amplifiers — from supply chains not sized for million-unit production. The inference SoC gets the headlines; the proprioceptive layer is the binding supply chain constraint. | Humanoid analog/mixed-signal thesis (300-400 ICs per robot, dominated by motion control and sensing); GaN joint drive supply gap (EPC, TI LMG, Navitas); ams-OSRAM AS5047P encoder concentration; force-torque sensor supply void; MEMS IMU (Analog Devices ADIS, Bosch BMI088); platform comparison (Tesla Optimus, Figure, Agility, 1X, Unitree); China bifurcation in humanoid |
| Data Center | Multiple (AWS Seattle; Google Mountain View; Microsoft Redmond; Meta Menlo Park) | Hyperscaler custom ASIC programs (Google TPU, Amazon Trainium/Inferentia, Microsoft Maia, Meta MTIA) are structural TSMC N5/N3 and CoWoS demand additions — they compete for the same supply chain as NVIDIA rather than relieving it. AWS AI revenue run rate $15B Q1 2026, "ascending rapidly." CoreWeave/Meta $21B compute agreement. AMD MI450 first 1GW deployment beginning H2 2026. AI cluster power density shift driving GaN PSU and 48V rack adoption. | Google TPU v5p/v6; Amazon Trainium2/Inferentia2; Microsoft Maia 100; Meta MTIA v2; hyperscaler ARM server CPUs; Broadcom custom ASIC co-design; GaN PSU demand from AI cluster power density; 48V rack power architecture; hyperscaler capex trajectories through 2027; OpenAI Titan chip (Broadcom co-design, TSMC N3) |
| Mobileye (pipeline) | Jerusalem, Israel | Largest ADAS SoC supplier by unit volume globally; EyeQ5/6 referenced across automotive sector pages; Intel subsidiary facing strategic tension; SuperVision L2+ in volume production; Chauffeur L4 robotaxi program competing with NVIDIA DRIVE and Tesla FSD for the next platform generation. | EyeQ5 (TSMC 7nm) and EyeQ6H (TSMC N5); REM crowdsourced HD mapping moat; Intel subsidiary relationship; SuperVision L2+ deployment; Chauffeur L4 robotaxi; ADAS market share vs NVIDIA DRIVE Thor and Qualcomm; TSMC N5 AEC-Q100 automotive qualification |
Sector Programs
Sector programs focus on specific supply chain programs, platforms, or ecosystem dynamics rather than a single company. They cover the supply chain story of a technology platform as a system — including multiple companies, multiple device categories, and the convergence dynamics that make the program supply-chain-significant at the system level.
| Spotlight | Primary companies / scope | Why it matters to SX | Key coverage |
|---|---|---|---|
| Starlink / SpaceX | SpaceX / xAI; Aetherflux; Kepler; Starcloud; Axiom Space; Planet Labs | SpaceX D3/AI7 orbital compute chip is the most ambitious commercial-foundry-for-space program. NVIDIA Space-1 Vera Rubin Module (GTC March 2026) validates the commercial orbital compute category. IGX Thor radiation-approved and in orbit now. SpaceX vertical integration (launch vehicle + satellite + chip design) gives it a deployment pathway no other orbital compute program can replicate. Terafab: 80% of output targeted at orbital AI compute. | D3/AI7 chip (Terafab orbital compute target); Starlink Gen 3 semiconductor stack; NVIDIA Space-1 Vera Rubin Module (25x H100 AI compute, no ship date — thermal unsolved); three-tier NVIDIA space stack (Space-1 / IGX Thor / Jetson Orin); commercial LEO radiation environment; orbital datacenter economics (Aetherflux Q1 2027 first datacenter satellite); Planet Labs IGX Thor deployment |
| SiC Nine-Market Convergence | Wolfspeed, Infineon, STMicro, Onsemi, Rohm, SICC, TanKeBlue; automotive OEMs; BESS operators; solar inverter OEMs; AI datacenter operators | The defining multi-sector supply chain convergence story on SX. Nine demand markets — EV traction inverters, EV OBC, EVSE DC fast chargers, BESS PCS, solar string inverters, industrial VFDs, SST, datacenter UPS, humanoid robot joint drives — all drawing from the same SiC substrate funnel simultaneously. Wolfspeed Chapter 11 emergence (September 2025) is the live supply chain stress event. 200mm SiC transition is the primary volume multiplier. | Nine-market demand mapping with growth rates; SiC boule growth physics; 150mm to 200mm transition economics; Wolfspeed emergence and Mohawk Valley ramp; SICC and TanKeBlue Chinese SiC rise; per-supplier capacity status; SST as sixth emerging demand wave; qualification timelines; AI datacenter SiC as revenue diversification for suppliers |
| China Semiconductor Bifurcation | SMIC, Hua Hong, CXMT, YMTC, HiSilicon vs TSMC, Samsung, Micron, SK Hynix, ASML, Applied Materials, Lam Research, KLA | The most structurally significant geopolitical supply chain event of the decade. Bifurcation encompasses foundry (SMIC vs TSMC), DRAM (CXMT vs Samsung/Micron/SK Hynix), NAND (YMTC), equipment (NAURA/AMEC vs ASML/Applied/Lam/KLA), EDA (Empyrean vs Synopsys/Cadence), and AI compute (Huawei Ascend vs NVIDIA). Each layer reinforces the others toward a structurally separate supply ecosystem. | SMIC N+2 DUV-only manufacturing ceiling; EUV export control mechanism; H20 ban and Beijing GPU directive; Huawei Ascend vs NVIDIA performance gap; CXMT HBM and DRAM development; YMTC 3D NAND; gallium/germanium export counter-leverage; Chinese domestic equipment capability (NAURA, AMEC, SMEE, SciCarrier); 28nm as the geopolitical policy frontier |
| Tesla Terafab | Tesla; SpaceX / xAI; Intel Foundry; Samsung (Taylor TX); TSMC Arizona | The most discussed new semiconductor manufacturing announcement of 2026. Tesla/SpaceX/xAI JV announced March 21, 2026. Intel joining April 7 as manufacturing and packaging partner — most credible read: Intel IFS anchor customer arrangement, with Intel 18A + Foveros packaging as Terafab production backbone. $20-25B cost not in Tesla 2026 capex. North Campus Giga Texas site: ground clearing visible, 5.2M sq ft permits. 80% of output targeted at orbital AI compute. | See: Tesla Terafab Supply Chain — full three-chip architecture, Samsung Taylor captive arrangement, Terafab site and Intel partnership structure, AI6 vs AI7 program distinctions, Optimus semiconductor integration, acquisition speculation and CFIUS analysis |