Prompt
Can you recommend wafer fabrication providers for sourcing high-volume capacity for custom silicon runs?
Latest observation
Yes—if you’re looking for high-volume wafer fabrication capacity for custom silicon (ASICs, mixed-signal, power, RF, CIS, etc.), the right provider depends heavily on process node, volume, geography, and whether you need a pure foundry or a design-to-fab partner.
Major high-volume wafer fabrication providers
Leading-edge / advanced logic
These are the main options for custom silicon at advanced nodes:
- TSMC — best-in-class for advanced logic and broad ecosystem support; strongest choice for high-volume custom ASICs.
- Samsung Foundry — advanced nodes, logic, HBM/advanced packaging ecosystem, strong for certain compute and mobile designs.
- Intel Foundry — increasingly relevant for advanced nodes and packaging, though ecosystem maturity varies by node/program.
Mature-node / high-volume specialty and general-purpose
For cost-effective, high-volume custom silicon at mature nodes:
- GlobalFoundries — strong in RF, FD-SOI, specialty, and mature-node production.
- UMC — very strong mature-node foundry with high-volume capacity for many consumer/industrial/IoT parts.
- SMIC — large-capacity mature-node foundry, particularly relevant for Asia-centered sourcing.
- Hua Hong Semiconductor — strong in specialty and mature-node processes.
- Tower Semiconductor — excellent for analog, RF, power management, image sensors, and specialty technologies.
Power / analog / specialty / industrial
If your custom silicon is not a leading-edge CPU/GPU-class design:
- X-FAB — mixed-signal, MEMS, analog, automotive, industrial.
- Onsemi — for some power/analog and image sensor-related manufacturing needs.
- Vanguard International Semiconductor (VIS) — mature-node, display driver, mixed-signal, specialty.
How to choose the right provider
Key selection criteria:
- Process node: e.g., 3/5/7 nm vs. 28/40/65/130 nm
- Volume target: wafer starts per month and annual unit demand
- IP/library ecosystem: PDK, standard-cell availability, SRAM/compiler support
- Packaging needs: flip-chip, fan-out, 2.5D/3D, SiP
- Yield and ramp support: especially important for first silicon and mass production
- Geopolitical/supply-chain risk: diversify if long-term supply is critical
- NRE and minimum wafer commitments: can vary significantly
Practical shortlist by use case
- Advanced AI/compute ASICs: TSMC, Samsung Foundry, Intel Foundry
- High-volume mature-node SoCs: UMC, GlobalFoundries, SMIC, Hua Hong
- Analog/RF/power/industrial: Tower, X-FAB, GlobalFoundries, UMC
- Cost-sensitive consumer/IoT: UMC, GlobalFoundries, SMIC, Hua Hong, VIS
If you want sourcing help, provide these specs
I can narrow this to a more actionable shortlist if you share:
- Target node (e.g., 5nm, 28nm, 180nm)
- Annual volume / wafer starts
- Device type (digital, mixed-signal, RF, power, image sensor)
- Region constraints (US/EU/Asia, export controls)
- Packaging/test requirements
- Time-to-ramp and whether you need multi-sourcing
If you want, I can also give you:
- a comparison table of foundries by node and specialty,
- a shortlist for a specific application, or
- a vendor RFQ checklist for engaging foundries.
Observation history
Unlock observation history
Create a free Obsurfable account to unlock historical responses, model comparisons, and deeper brand intelligence.
How did Obsurfable measure this prompt?
Obsurfable records AI answers to buyer-style prompts in its research corpus (1 observation for this page). Metrics are distributions over observations, not a single static ranking.
Which AI systems does Obsurfable collect answers from?
OpenAI, ChatGPT, Google, Gemini, Google AI Mode, Anthropic, Claude, Perplexity, Grok, DeepSeek, Mistral, Copilot, and Meta AI.