East Asian Technology Intelligence
Japan & China technology, translated and contextualized for Western readers
A Taiwan-based ASIC design-service company that designs and manages production of custom AI and high-performance-computing processors for cloud providers and system builders.
Alchip occupies a valuable but externally dependent position in custom AI silicon. It benefits sharply from a small number of high-value accelerator programs, but this project-based model produces material revenue volatility, concentration risk, and dependence on external foundry and packaging capacity.
Alchip Technologies is a Taiwan-based ASIC design-service company founded in 2003. It earns revenue by designing and managing the production of custom chips, particularly advanced-node AI and high-performance-computing processors, rather than by operating its own semiconductor fab. The company supplies architecture support, physical design, tape-out execution, manufacturing management, advanced-package design, chiplet integration, and design work for 2.5D/3D integration.
Alchip converts customer accelerator specifications into advanced-node silicon and production-ready packaged chips for hyperscaler and HPC deployment. It depends heavily on external foundry capacity and process-design ecosystems, particularly advanced-node manufacturing and advanced packaging. Its reported AI-related capabilities include CoWoS, Integrated Fan-Out, 3DIC, multi-die architectures, and advanced-node designs from 7 nm through 2 nm. The company has completed more than 600 designs through the end of 2025.
AI demand reaches Alchip directly through custom accelerator programs. The company designs complex ASICs and manages their manufacturing at advanced nodes, including 7 nm, 5 nm, 3 nm, and 2 nm designs, for cloud service providers and edge-AI applications. Its revenue mechanism is primarily non-recurring engineering during the design and tape-out phases, followed by production-related revenue when customer chips enter volume manufacturing.
In FY2025, AI and HPC applications accounted for 83% of total revenue, while designs at 7 nm and below accounted for 87%. The company attributes a significant Q2 2026 sequential revenue increase to the ramp of a 3 nm AI accelerator for a North American customer, which alone accounted for 51% of quarterly revenue. In that same quarter, 3 nm and 2 nm designs accounted for 47% of quarterly revenue, while 5 nm and 7 nm designs accounted for another 40%. Alchip's AI exposure is commercial and direct, converting hyperscaler specifications into deployed silicon without commercializing an accelerator of its own.
Alchip's competitive position rests on advanced ASIC implementation, manufacturing management, and package-aware design work. The company reports completed 3 nm tape-outs, a 2 nm test-chip tape-out in 2024, several 2 nm high-performance design projects in 2025, and designs using CoWoS and Integrated Fan-Out packaging.
The technical challenge is not simply RTL or standard-cell design. Advanced accelerator programs require physical design that closes timing, power, signal integrity, thermals, memory interfaces, die-to-die connections, package rules, yield requirements, and foundry design-rule constraints. Once an ASIC reaches tape-out or begins volume production, switching design houses can delay a program because the replacement must absorb customer IP, revalidate implementation decisions, and regain foundry and packaging confidence. Alchip's primary constraint is scale: with 645 employees, it has less redundancy and a smaller IP library than global EDA or large merchant-chip competitors, making execution and customer trust critical.
Alchip depends heavily on external foundry capacity and process-design ecosystems, particularly advanced-node manufacturing and advanced packaging. While it does not name a principal foundry in its FY2025 annual report, TSMC dependence is inferred from Alchip's advanced-node and CoWoS-related service model, as well as historical equity investments and the presence of a former TSMC advanced-packaging director on its board.
On the customer side, Alchip does not disclose the names of its largest chip-design customers. It describes active programs for cloud-service-provider AI accelerators, edge-AI chips, high-performance computing, networking, consumer applications, and automotive ADAS. A single unnamed North American cloud-service-provider customer accounted for 51% of Q2 2026 revenue during a 3 nm accelerator ramp, highlighting extreme customer concentration.
Alchip's headquarters and central engineering presence are in Taipei, exposing it to Taiwan cross-strait disruption and reliance on Taiwan's advanced semiconductor ecosystem. A disruption to Taiwan operations or foundry logistics would affect design support and customer production schedules.
The company also faces export-control exposure because AI accelerators and advanced-node custom ASICs can fall within US controls on technology, semiconductor design tools, advanced manufacturing, and China end-use. Alchip maintains subsidiaries in mainland China, including Shanghai, Wuxi, Hefei, Jinan, Guangzhou, and Chongqing. China-related controls or restrictions could complicate staffing, customer work, IP handling, or cross-border technology transfer. The company also operates subsidiaries in the United States, Japan, Malaysia, and Vietnam.
| Risk | Severity | Why it matters |
|---|---|---|
| Taiwan cross-strait disruption | High | Headquarters and central engineering rely on Taiwan's advanced semiconductor ecosystem. |
| Advanced foundry dependence | High | Relies entirely on external manufacturing and capacity allocation for 3 nm, 2 nm, and CoWoS programs. |
| Export-control exposure | High | Advanced AI ASICs fall within US controls on technology and China end-use. |
| Customer concentration | High | A single North American customer accounts for 51% of Q2 2026 revenue. |
| Program-timing volatility | High | Revenue model is exposed to tape-out timing and the handoff from NRE to production. |
| China operational exposure | Medium | Maintains multiple subsidiaries in mainland China, complicating IP handling and staffing. |
| Engineering-talent retention | Medium | A 645-person organization has less redundancy than global EDA or large merchant-chip competitors. |
| Margin-mix pressure | Medium | AI-volume ramps can improve revenue while reducing near-term percentage margins. |
Alchip is an independent public company incorporated in the Cayman Islands and listed on the Taiwan Stock Exchange. It is not state-owned or controlled by a chaebol-style industrial group, and its FY2025 governance report identifies no state shareholder or controlling corporate parent.
The board consists of seven directors, four of whom are independent, representing approximately 57% of the board. Johnny Shyang-Lin Shen serves as both chairman and CEO, holding a 2.30% stake as of March 2026. The company's FY2025 annual report is audited by Deloitte & Touche, indicating dispersed public ownership and standard professional governance, though the combined chairman and CEO role concentrates executive leadership.
We haven't written about Alchip yet. Subscribe below to get it when we do.
Compiled with AI-assisted research from company filings, market data, and published reporting as of September 13, 2026, then reviewed by AsiaAI.FYI. Figures marked Estimate are not company-reported. Check primary filings before relying on any number.
Confidence: B. Financial results, management, and core technical claims are supported by audited filings, but customer identities, supplier terms, and patent counts are not disclosed.
Main sources: Alchip FY2025 annual report; Alchip FY2024 annual report and audited consolidated financial statements; Company financial-results releases; Taiwan and international market-data sources.
East Asian Technology Intelligence
Japan & China tech news — translated, contextualized, and delivered for Western readers.
Free. Unsubscribe anytime.