What it holds
AI Chips follows the money through the Semiconductor 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 |
How it breaks down
- Taiwan59%
- China26%
- United States15%
- Mega cap (over $200bn)30%
- Large cap ($10bn to $200bn)55%
- Mid cap ($2bn to $10bn)15%
Shares of the strategy as of the data date.
Largest companies
| Company | Listing | Market value |
|---|---|---|
| Nvidia | NVDA.XNAS | $5.2 trillion |
| Taiwan Semiconductor Manufacturing | 2330.TW | $1.9 trillion |
| ASE Technology Holding | 3711.TW | $95 billion |
| Hygon Information Technology Co. Ltd. A | 688041.SHG | $67 billion |
| Alchip Technologies | 3661.TW | $10 billion |
| Global Unichip | 3443.TW | $34 billion |
| Cambricon Technologies | 688256.SHG | $61 billion |
The market
All advanced packaging, from over USD 46 bn in 2024; includes TSMC CoWoS capacity coverage
Yole Group (Advanced Packaging Market Monitor), 2030Pure-play foundry sector full-year growth, driven by AI GPUs, ASICs and accelerators on 4/5 nm (a year)
Counterpoint Research, 2025CAGR for all advanced packaging (growth, a year)
Yole Group, 2024 to 2030AI accelerator forecast, raised from the earlier USD 500 bn 2028 figure
AMD (Lisa Su, AMD Advancing AI 2026, as reported by Futurum), 2030Whole-segment AI chip 2024 estimates range from USD 21 bn to USD 118 bn, and data center GPU estimates from USD 3.2 bn to USD 36 bn, mostly because scope definitions differ. The AMD and JPMorgan figures cover only data center accelerators or custom ASICs, and the foundry and packaging figures cover all leading-edge logic or all advanced packaging rather than AI chips alone.
All 28 published estimates, with sources
| What it measures | Figure | Year | Source |
|---|---|---|---|
| AI chips; incremental market growth over the period, not total market size | $155bn | 2025 to 2030 | Technavio |
| AI chips; rising to about USD 293 bn by 2030; growth rate not read | $118bn | 2024 | MarkNtel Advisors |
| AI chips; rising to USD 118.05 bn by 2030 | $21bn | 2024 | TechSci Research |
| AI chips; rising to USD 295.56 bn by 2030 | $53bn | 2024 | Next Move Strategy Consulting |
| AI chipsets: CPU, GPU, FPGA, ASIC and others; rising to USD 323.14 bn by 2030 | $57bn | 2023 | Grand View Research |
| CAGR for AI chips (growth, a year) | +33% | 2025 to 2030 | TechSci Research |
| CAGR for AI chips (growth, a year) | +24% | 2025 to 2030 | Technavio |
| CAGR for AI chips; period assumed to end 2030 from the 2030 figure (growth, a year) | +33% | 2024 to 2030 | Next Move Strategy Consulting |
| CAGR for AI chipsets, CPU, GPU, FPGA, ASIC and others (growth, a year) | +29% | 2023 to 2030 | Grand View Research |
| All advanced packaging, from over USD 46 bn in 2024; includes TSMC CoWoS capacity coverage | $80bn | 2030 | Yole Group (Advanced Packaging Market Monitor) |
| All advanced packaging; up 19% YoY; projected to exceed USD 79.4 bn by 2030 | $46bn | 2024 | Yole Group |
| CAGR for all advanced packaging (growth, a year) | +9.5% | 2024 to 2030 | Yole Group |
| CAGR for all advanced packaging (growth, a year) | +9.6% | 2024 to 2030 | Yole Group (Advanced Packaging Market Monitor) |
| CAGR for nodes at 7 nm and below (all leading-edge logic) (growth, a year) | +13% | 2025 to 2035 | GM Insights |
| Nodes at 7 nm and below (all leading-edge logic, not only AI chips); rising to USD 150.30 bn by 2035 | $43bn | 2025 | GM Insights |
| Pure-play foundry sector full-year growth, driven by AI GPUs, ASICs and accelerators on 4/5 nm (a year) | +26% | 2025 | Counterpoint Research |
| AI accelerator forecast, raised from the earlier USD 500 bn 2028 figure | $1.4 trillion | 2030 | AMD (Lisa Su, AMD Advancing AI 2026, as reported by Futurum) |
| CAGR for data center GPUs (growth, a year) | +31% | 2024 to 2030 | Research and Markets (via Yahoo Finance) |
| Communications and infrastructure segment of advanced packaging, which includes AI accelerators and GPUs (growth, a year) | +15% | 2024 to 2030 | Yole Group |
| Custom AI ASIC market CAGR; range 40–50%; growth period not given (a year) | +40% | not stated | JPMorgan (as reported by CiVL) |
| Custom AI ASIC market; range USD 60–70 bn in 2026 | $60bn | 2026 | JPMorgan (as reported by CiVL) |
| Data center AI accelerator market 2023 base, as stated in a social media post | $45bn | 2023 | X post (EconomyApp), citing AMD base |
| Data center AI accelerator market, including GPUs, FPGAs and other accelerators | $500bn | 2028 | AMD (Lisa Su, Q3 2024 earnings call, as reported by Yahoo Finance) |
| Data center GPUs; reaching US$92.0 bn by 2030 | $18bn | 2024 | Research and Markets (via Yahoo Finance) |
| Data center GPUs; reaching USD 183 bn by 2034; the page's two growth rates conflict, so no CAGR is recorded | $18bn | 2024 | Market.us |
| Data center GPUs; reaching USD 22.46 bn by 2034 (low outlier) | $3.2bn | 2024 | Research and Markets (via Yahoo Finance) |
| Data center GPUs; reaching USD 293.2 bn by 2034 (high outlier) | $36bn | 2024 | BIS Research |
| Growth of the data center AI accelerator market, stated as more than 60% a year (a year) | +60% | to 2028 | AMD (Lisa Su, Q3 2024 earnings call, as reported by Yahoo Finance) |
Questions
- What is AI Chips?
- GPUs, custom accelerators, leading-edge foundry and advanced packaging.
- What does AI Chips hold?
- Listed companies that earn their revenue from the parts it takes to build AI Supercycle: Graphics Processing Unit (GPU). 7 companies as of 2026-10-28.
- What is not in AI Chips?
- NVLink and other scale up switch chips: Interconnection Interface, not Graphics Processing Unit (GPU) 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)
- How large is the AI Chips market?
- About 56.8 USD billion (2023), the middle of independent estimates that agree (for example Grand View Research).
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: io.net/p/gb300-nvl72-cloud-rental-how-to-access-nvidias-blac · panews.io/articles/01a1259f-d881-7008-9a96-71832906cd45 · bitget.com/news/detail/12560605422208 · datagravity.dev/p/how-much-does-an-nvidia-nvl72-cost · investing.com/news/stock-market-news/how-much-does-a-gw-of-d · investing.com/news/technology-news/what-it-costs-to-deploy-1 · itiger.com/news/1113387068 · remio.ai/post/bernstein-ai-data-center-costs-hit-39-5-billio · spheron.network/blog/gb300-nvl72-vs-gb200-nvl72-pricing-avai · tomshardware.com/pc-components/gpus/nvidias-next-gen-blackwe · x.com/jarrenfeldman/status/1942370046606729367 · x.com/pequityresearch/status/2084631439149006849
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.
- NVLink and other scale up switch chips: Interconnection Interface, not Graphics Processing Unit (GPU)
- 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)
- CPUs including those bundled in GPU superchip modules: Servers and IT Equipment, not Graphics Processing Unit (GPU)
Not investment advice. Figures are dated and sourced; ThematIQ does not estimate figures it cannot source.