What it holds
AI Semiconductors follows the money through the AI bill of materials: the parts it takes to build it, and the companies that earn their revenue supplying them. Each company is read from its own reports, so a holding reflects what it actually sells into the theme, not what it says about it.
| Part | What it covers |
|---|---|
| Graphics Processing Unit (GPU) | GPU accelerators, AI ASICs custom, foundry leading edge, advanced packaging |
| Video Memory (VRAM) | HBM memory, GDDR memory, DRAM |
| Data Processing Unit (DPU) | DPU SmartNIC |
| Cooling System | direct to chip cooling, thermal management components |
| Voltage Regulator Module (VRM) | voltage regulator modules, power management ICs |
| Motherboard Interface | server motherboards, PCIe CXL retimers switches |
| Interconnection Interface | infiniband, ethernet switch silicon, optical transceivers, SerDes ip, interconnect ICs |
| Network Interface and Controller | NIC controllers |
| Output Interfaces | display interface ICs |
| Software (AI hardware performance) | AI hardware software stack, EDA software |
How it breaks down
- United States43%
- Taiwan31%
- China22%
- South Korea4%
- Mega cap (over $200bn)29%
- Large cap ($10bn to $200bn)63%
- Mid cap ($2bn to $10bn)8%
- Small cap (under $2bn)1%
Shares of the strategy as of the data date.
Largest companies
| Company | Listing | Market value |
|---|---|---|
| Micron Technology | MU.XNAS | $1.1 trillion |
| Coherent | COHR.XNYS | $47 billion |
| Nvidia | NVDA.XNAS | $5.2 trillion |
| Taiwan Semiconductor Manufacturing | 2330.TW | $1.9 trillion |
| ASE Technology Holding | 3711.TW | $95 billion |
| Fabrinet | FN.XNYS | $13 billion |
| Lumentum Holdings | LITE.XNAS | $82 billion |
| Marvell Technology | MRVL.XNAS | $221 billion |
| SK Hynix | 000660.KO | $786 billion |
| Realtek Semiconductor | 2379.TW | $13 billion |
| Shandong Zhongji Electrical Equipment | 300308.SHE | $106 billion |
| Eoptolink Technology | 300502.SHE | $38 billion |
| Global Unichip | 3443.TW | $34 billion |
| Alchip Technologies | 3661.TW | $10 billion |
| Hygon Information Technology Co. Ltd. A | 688041.SHG | $67 billion |
The market
Total semiconductor market, wider than AI
WSTS (trade body, Spring 2026 forecast), 2026HBM total addressable market
Micron (company earnings call, 17 December 2025), 2028AI HBM market (growth, a year)
SK hynix (as reported by Reuters and others, August 2025), to 2030AI chips, mostly GPUs, ASICs and accelerators
Next Move Strategy Consulting, 2024Whole-segment estimates range from USD 4 billion to USD 53 billion for 2024 because vendors define AI chips differently, while company and trade-body part-level figures such as Micron's HBM TAM are far larger or use different scopes, so the figures are not directly comparable.
All 20 published estimates, with sources
| What it measures | Figure | Year | Source |
|---|---|---|---|
| AI chips (growth of USD 154.93 billion over the period) (a year) | +24% | 2025-2030 | Technavio |
| AI chips (whole segment estimate) (growth, a year) | +33% | 2025-2030 | TechSci Research |
| AI chips (whole segment estimate) | $21bn | 2024 | TechSci Research |
| AI chips, mostly GPUs, ASICs and accelerators (growth, a year) | +33% | 2024-2030 | Next Move Strategy Consulting |
| AI chips, mostly GPUs, ASICs and accelerators | $53bn | 2024 | Next Move Strategy Consulting |
| AI chipset market | $4.0bn | 2024 | Market Research Future |
| AI data-centre liquid cooling (direct-to-chip and immersion) (growth, a year) | +34% | 2026-2032 | MarketsandMarkets |
| AI data-centre liquid cooling (direct-to-chip and immersion) | $3.5bn | 2026 | MarketsandMarkets |
| Total semiconductor market, wider than AI | $1.5 trillion | 2026 | WSTS (trade body, Spring 2026 forecast) |
| AI data-centre DPU and SmartNIC only (growth, a year) | +28% | 2026-2032 | Knowledge Sourcing Intelligence |
| AI data-centre DPU and SmartNIC only | $1.9bn | 2026 | Knowledge Sourcing Intelligence |
| AI HBM market (growth, a year) | +30% | to 2030 | SK hynix (as reported by Reuters and others, August 2025) |
| DPU and SmartNIC market (growth, a year) | +15% | 2025-2034 | Polaris Market Research |
| DPU and SmartNIC market | $1.1bn | 2024 | Polaris Market Research |
| HBM total addressable market (growth, a year) | +40% | 2025-2028 | Micron (company earnings call, 17 December 2025) |
| HBM total addressable market | $35bn | 2025 | Micron (company earnings call, 17 December 2025) |
| HBM total addressable market | $100bn | 2028 | Micron (company earnings call, 17 December 2025) |
| High-bandwidth memory (HBM) | $3.4bn | 2025 | Market Research Future |
| Optical transceivers and related products (growth, a year) | +55% | 2025 | LightCounting |
| Optical transceivers and related products | $24bn | 2025 | LightCounting |
Questions
- What is AI Semiconductors?
- GPUs and accelerators, memory, chip power, interconnect and the chip software stack.
- What does AI Semiconductors hold?
- Listed companies that earn their revenue from the parts it takes to build AI Supercycle: Graphics Processing Unit (GPU), Video Memory (VRAM), Data Processing Unit (DPU), Cooling System, Voltage Regulator Module (VRM), Motherboard Interface. 35 companies as of 2026-10-28.
- What is not in AI Semiconductors?
- Cold plates, vapor chambers, thermal interface materials and server level liquid loops: Cooling System, not Power and Cooling Systems; CDUs, rack manifolds, rear door exchangers, chillers and CRAH units: Power and Cooling Systems, not Cooling System Standard and RDMA NIC adapters and controllers: Network Interface and Controller, not Networking Equipment; programmable SmartNICs and DPUs with onboard compute cores: Data Processing Unit (DPU), not Network Interface and Controller Pluggable optical transceivers, AOCs, DACs and AECs: Interconnection Interface, not Networking Equipment or Racks and Cable Management; long haul and DCI optical transport systems: Networking Equipment; passive fiber trunks, patch panels and structured cabling: Racks and Cable Management
Sources
Every part of the bill of materials is researched from published industry sources, and every company from its own annual reports and exchange data. Nothing is estimated that cannot be sourced.
Industry research: ai-stack.us/products/2-248-nvidia-connectx-8-supernic-900-9x · blog.barrack.ai/nvidia-b300-1400w-data-center-requirements/ · capitalandcompute.net/memory-prices/hbm/ · dev.to/gentic_news/nvidia-feynman-gpu-power-semi-content-hit · docs.nvidia.com/enterprise-reference-architectures/nvl72-ai- · epoch.ai/data-insights/b200-cost-breakdown · finance.biggo.com/news/X4vnUp4B2jrwCtglf11J · finance.yahoo.com/news/cooling-system-single-nvidia-blackwel · finance.yahoo.com/technology/ai/articles/nvidia-already-make · fourweekmba.com/ai-nvidias-huang-pegs-software-at-about-2000 · fourweekmba.com/hbm-the-memory-that-controls-45-of-gpu-econo · gpuservercase.com/blog/display-output-in-gpu-servers-a-buyer · intuitionlabs.ai/articles/hbm-dram-ai-memory-demand · io.net/p/gb300-nvl72-cloud-rental-how-to-access-nvidias-blac · jpthoughtsandgrowth.substack.com/p/from-substation-to-chip-a · lenovopress.lenovo.com/lp2357-lenovo-nvidia-gb300-nvl72-rack · mixed-news.com/en/nvidia-huang-2000-gpu-software-licensing-e · newsletter.semianalysis.com/p/co-packaged-optics-cpo-book-sc · panews.io/articles/01a1259f-d881-7008-9a96-71832906cd45 · qsfpdd800g.com/blogs/artical/1-6t-800g-osfp224-optical-trans
What it is not
Where an item could belong to two parts, these rules decide which owns it; items outside every part are not counted.
- Cold plates, vapor chambers, thermal interface materials and server level liquid loops: Cooling System, not Power and Cooling Systems; CDUs, rack manifolds, rear door exchangers, chillers and CRAH units: Power and Cooling Systems, not Cooling System
- Standard and RDMA NIC adapters and controllers: Network Interface and Controller, not Networking Equipment; programmable SmartNICs and DPUs with onboard compute cores: Data Processing Unit (DPU), not Network Interface and Controller
- Pluggable optical transceivers, AOCs, DACs and AECs: Interconnection Interface, not Networking Equipment or Racks and Cable Management; long haul and DCI optical transport systems: Networking Equipment; passive fiber trunks, patch panels and structured cabling: Racks and Cable Management
- Finished Ethernet and InfiniBand switch systems: Networking Equipment, not Interconnection Interface; merchant switch ASICs and InfiniBand silicon sold as chips: Interconnection Interface, not Networking Equipment
- Board level VRMs, power stages and PMICs: Voltage Regulator Module (VRM), not Power and Cooling Systems
- Server PSUs, rack power shelves and busbars: Power and Cooling Systems, not Servers and IT Equipment, Voltage Regulator Module (VRM) or Racks and Cable Management; power semiconductors inside them counted only within the PSU price
- Server motherboards and GPU baseboards: Motherboard Interface, not Servers and IT Equipment
- ODM rack scale assembly and test value: Servers and IT Equipment, not Racks and Cable Management or Motherboard Interface; rack enclosure hardware itself: Racks and Cable Management
- High speed copper backplanes, scale up cable cartridges and high speed connectors: Interconnection Interface, not Racks and Cable Management
- NVLink and other scale up switch chips: Interconnection Interface, not Graphics Processing Unit (GPU)
- PCIe and CXL retimers and switches: Motherboard Interface, not Interconnection Interface; Ethernet and optical DSPs and standalone SerDes IP: Interconnection Interface
- Custom accelerator design services and IP embedded in AI ASICs: Graphics Processing Unit (GPU), not Interconnection Interface; standalone SerDes IP licensing: Interconnection Interface
- HBM stacks including TSV stacking and base die: Video Memory (VRAM), not Graphics Processing Unit (GPU); interposer and 2.5D integration: Graphics Processing Unit (GPU)
- Server DRAM modules: Video Memory (VRAM), not Servers and IT Equipment
- CPUs including those bundled in GPU superchip modules: Servers and IT Equipment, not Graphics Processing Unit (GPU)
- General purpose server OS, virtualization and cluster orchestration: Datacenter Software, not Software (AI hardware performance); GPU drivers, CUDA, ROCm, AI compilers and inference runtimes: Software (AI hardware performance)
- SSD controllers and RAID or NVMe Storage Devices controllers: Storage Devices, not Data Processing Unit (DPU) or Motherboard Interface
Not investment advice. Figures are dated and sourced; ThematIQ does not estimate figures it cannot source.