Taiwan AI-Server ODMs: Quarterly Revenue (Q2 2026)
Hon Hai, Quanta, Wistron, Wiwynn and Inventec — Q2 2026 consolidated revenue and AI-server momentum for Taiwan's five big AI-server ODMs, the supply-chain core of the global AI compute buildout.
| Company (ticker) | Q2 2026 revenue | YoY | AI-server disclosure | Focus |
|---|---|---|---|---|
| Hon Hai / Foxconn (2317) | ~NT$2.53tn | +41% | Cloud & networking = 51% of revenue (first time >half; call basis) | World's largest EMS; iPhone assembly + AI-server GB200/GB300 racks |
| Quanta (2382) | ~NT$1.04tn | +105.6% | AI servers ~75–80% of total server revenue | AI-server systems/racks + notebook ODM |
| Wistron (3231) | NT$895.4bn | +62.4% | AI servers >50% of server revenue (call basis) | PC-ODM turned AI-server (GPU boards/L6) |
| Wiwynn (6669) | NT$278.2bn | +26.0% | Not broken out (business almost entirely cloud servers) | Cloud/hyperscale white-box servers (Meta, Microsoft) |
| Inventec (2356) | NT$269.9bn | +44.6% | Not broken out (Q2 driven by L6 boards; H2 L10/L11) | Server boards/L6 + notebooks; building L10/L11 racks & ASIC |
Method & sources
Consolidated revenue and YoY for Q2 2026 (Apr 1 – Jun 30, 2026), retrieved 2026-08-19. **First-party (official press release / IR; revenue and EPS trace verbatim to the official page, precise to hundreds of millions)**: Hon Hai https://www.foxconn.com/en-us/press-center/press-releases/latest-news/2095 ; Wistron https://www.wistron.com/en/Newsroom/2026-08-04 ; Wiwynn https://www.wiwynn.com/news/wiwynn-reports-second-quarter-2026-financial-results . **Media-relayed only (both firms file via Taiwan's MOPS and issue no English press release)**: Quanta and Inventec quarterly figures come from the earnings calls, cross-checked across 3+ financial outlets and self-consistent with the prior quarter — Quanta https://www.chinatimes.com/realtimenews/20260813004224-260410 (revenue rounded to the trillion); Inventec https://ec.ltn.com.tw/article/breakingnews/5535919 (official finance page currently 404). **Standalone AI-server revenue is disclosed by none of the five**; the "AI-server disclosure" column lists only company-stated proportions or qualitative notes — Quanta "~75–80% of total server revenue", Hon Hai "cloud & networking (AI-server-led) = 51% of consolidated revenue", Wistron ">50% of server revenue" (earnings-call basis). **Do not back out absolute AI-server figures from these ratios.** From April 2026 Wiwynn moved some customers' memory to a consignment (agency) model, excluded from revenue and cost, which shifts the YoY comparison basis.
Source: https://www.foxconn.com/en-us/press-center/press-releases/latest-news/2095
Retrieved: 2026-08-19