White Box Server Companies

White Box Server Market Opening Overview

Why the White Box Server Market Is Expanding

The White Box Server Market is undergoing the fastest structural shift in server procurement history, moving away from branded OEM chassis toward specification-driven, ODM-built platforms designed under the Open Compute Project (OCP) framework. The market reached USD 22.48 Billion in 2025 and is projected to climb to USD 82.61 Billion by 2035, a 13.9% CAGR, as hyperscalers led by Meta, Microsoft, and Google redirect procurement toward original design manufacturers to cut per-unit costs by 20–30%. Rack-mounted servers held 38.6% share in 2025, but GPU servers are forecast to grow fastest at 15.5% CAGR as 700W-plus accelerators outpace what legacy OEM chassis were engineered to cool. 

Two forces are compounding this shift. First, GPU-dense AI training clusters — now exceeding 25,000 accelerators per deployment — demand thermal and power-density engineering that Taiwanese ODMs iterate on six-month cycles, versus the 18-month cadence typical of branded vendors. Second, sovereign-cloud mandates in the EU, India and the Gulf states are steering government and regulated-enterprise buyers toward cost-efficient, vendor-neutral white-box stacks. North America holds a 32.4% share of hyperscaler capex, while Asia-Pacific is the fastest-growing region at 16.3% CAGR, anchored by Taiwan's ODM corridor and India's data-center build-out.

What Structurally Separates Leaders From the Field

It's not just price that drives this industry -- it's the ability to co-design with silicon vendors pre-volume and factory-integrated liquid-cooling capacity that air-cooled competitors can't retrofit fast enough. Wiwynn, Quanta, Wistron and Foxconn are involved in the reference-architecture programs of both NVIDIA and AMD at the design stage, which enables them to supply qualifying racks within weeks of a new GPU introduction. AMD’s outright acquisition of ZT Systems’ design team and Sanmina’s subsequent purchase of its manufacturing arm signals that even chip vendors and EMS players now see direct rack-scale system ownership—not arm’s-length ODM contracts—as the structural advantage this market rewards.

Top 10 Global White Box Server Companies — MRFR Rankings (2026)

All revenue figures are validated from official company financial disclosures, investor relations announcements, or SEC/regulatory filings. Where official figures are unavailable for private companies, this is explicitly noted.

*Company revenue share ranking follows MRFR's Competitive Benchmarking analysis of estimated white-box server segment revenue, which differs from total consolidated company revenue for diversified manufacturers such as Foxconn, Compal, and Pegatron.

Detailed Company Profiles

1. Quanta Computer Inc.  |  TWSE: 2382  |  Taoyuan City, Taiwan

Quanta is the volume leader of the White Box Server Market, converting a legacy notebook-ODM franchise into the industry's deepest hyperscaler relationship base. FY2024 revenue reached NT$1.41 Trillion (Official Annual Audit Report), and January–September 2025 revenue hit a record NT$1.49 Trillion, up 49.5% YoY, as AI servers grew to roughly half of total revenue. Quanta holds early-access design status inside NVIDIA's HGX/MGX programs and is expanding production capacity in Taiwan, Thailand, and the US, with plans to at least double AI server output by end-2026. MRFR views Quanta's design-stage silicon access as a moat competitors cannot buy — only qualify into.

2. Wiwynn Corporation  |  TWSE: 6669  |  New Taipei City, Taiwan

Wiwynn, spun out of Wistron to focus purely on cloud infrastructure, posted record FY2024 consolidated revenue of NT$360.541 Billion (Official Q4 2024 Results), up 49.0% YoY, with margins also at record highs. The company is investing a further USD 300 Million to expand its U.S. subsidiary WYMUS in Texas and continues co-developing NVIDIA Blackwell Ultra-based AI servers with Wistron. Wiwynn's specialization in cold-plate and two-phase immersion cooling positions it to capture rack designs that air-cooled OEM competitors cannot yet deliver at scale.

3. Inventec Corporation  |  TWSE: 2356  |  Taipei, Taiwan

Inventec FY2024 Consolidated Revenue NT$646.2 Billion (Official Investor Relations Disclosure), up 25.6% YoY Led by Server/AI Demand and 2025 Revenue to Surpass NT$691.1 Billion. Inventec is one of four ODMs selected for NVIDIA’s first HGX Partner Program and provides rack, blade, and HPC nodes in Japan, China, and global OEM channels. Instead of single-customer focus, MRFR cites Inventec’s broad architecture coverage as its structural hedge against hyperscaler demand volatility.

4. MiTAC Holdings Corporation  |  TWSE: 3706  |  Taoyuan, Taiwan

MiTAC Holdings - FY2024 consolidated revenue of NT$61.360 Billion (Official Board-Approved Financial Report). 2025 revenue surges over 72% on accelerated demand for storage and edge servers. MiTAC, under the TYAN brand, has purposely stayed away from the largest hyperscale GPU-rack deals and instead secured a niche in storage-optimized and embedded server designs. This narrower focus limits its addressable ceiling, but insulates margins from the design-cycle price wars playing out among the top-five volume leaders.

5. Super Micro Computer, Inc.  |  NASDAQ: SMCI  |  San Jose, CA, USA

Supermicro is the sole leading white-box vendor based and vertically integrated in the United States, with net sales of USD 22.0 Billion (SEC Filing / Official Press Release) for FY2025 (ending June 30, 2025), up from USD 15.0 Billion in FY2024. Its factory-integrated direct-liquid-cooling racks provide it with an approximately six-month time-to-deployment advantage over retrofit approaches from slower-moving OEM rivals. Supermicro’s return to timely reporting and ongoing design-cycle speed, after a spell of overdue SEC filings that rocked investor trust, helps confirm MRFR’s opinion that US-based manufacturing scale is emerging as a distinct competitive lever, not a compliance need.

6. Foxconn (Hon Hai Precision Industry Co., Ltd.)  |  TWSE: 2317  |  New Taipei City, Taiwan

Hon Hai Technology Group posted FY2025 group revenue of NT$8.1 Trillion (Official FY2025 Results), up 18% YoY, though white-box servers represent only a portion of a business still anchored by consumer electronics assembly. Cloud-and-networking is now Foxconn's largest single segment, and the Group's AI server revenue is approaching trillion-NTD scale on a standalone basis. Foxconn's scale advantage lies in its 230+ campus manufacturing footprint spanning Taiwan, the US and Mexico — a logistics and geopolitical-diversification asset that narrower ODM peers cannot replicate at comparable cost.

7. Compal Electronics, Inc.  |  TAIEX: 2324  |  Taipei, Taiwan

Compal reported FY2024 consolidated revenue of NT$910.253 Billion (Official Board-Approved Financial Report), down 4% YoY as legacy notebook demand softened, even as operating profit rose 23%. The company is deliberately re-weighting toward non-PC revenue — including servers — targeting 40% of sales by 2026, up from 25% in 2024. MRFR reads Compal's margin improvement despite falling top-line revenue as evidence that its value-tier server pivot, not volume alone, is now the more important signal to track.

8. Pegatron Corporation  |  TWSE: 4938  |  Taipei, Taiwan

Pegatron's consolidated revenue reached NT$1,117.2 Billion in FY2025 (Official Annual Report), roughly flat year-on-year, with server shipments guided to more than double quarter-over-quarter into Q1 2026 as AI server demand offsets softness in consumer electronics. Pegatron continues platform-level R&D across NVIDIA, AMD and Intel server references, leveraging manufacturing scale built for consumer electronics to absorb server-segment growth without proportionate capital outlay. Its position remains that of a fast-follower rather than a design-stage partner on the newest GPU architectures.

9. ZT Systems  |  Private (design unit retained by AMD; manufacturing acquired by Sanmina)  |  Secaucus, NJ, USA

ZT Systems generated an estimated USD 10 Billion in annualized revenue at the time AMD announced its USD 4.9 Billion acquisition in August 2024 (Official AMD Press Release) — the company itself remains private with no independently published financial statements. AMD retained ZT Systems' AI rack-scale design team, while its data-center manufacturing business was sold to Sanmina for up to USD 3 Billion in a deal that closed in October 2025. MRFR views this split — chip vendor absorbing design, EMS player absorbing manufacturing — as a template competitors will likely replicate as silicon vendors seek tighter control over rack-level system architecture.

10. Celestica Inc.  |  TSX/NYSE: CLS  |  Toronto, Canada

Celestica delivered its strongest financial year on record in FY2024, with revenue of USD 9.65 Billion (SEC Annual Report to Shareholders), up 21% YoY, and GAAP EPS up 78% to USD 3.61. Growth is concentrated in its Hardware Platform Solutions (HPS) segment, which serves hyperscale networking and server customers alongside Celestica's legacy EMS business. MRFR notes Celestica's diversification beyond pure-play server manufacturing into networking integration as a buffer against the single-segment cyclicality that pure ODM competitors face.

M&A Activity Tracker

Key verified transactions and strategic investments shaping White Box Server Market consolidation (2024–2026):

Key Trend: The AMD–ZT Systems–Sanmina sequence signals a structural split between AI rack design (increasingly absorbed by chip vendors and hyperscalers) and manufacturing scale (increasingly consolidated among EMS players), while Wistron's and Nvidia's Texas capacity build-outs show geography, not just ownership, is now a competitive lever in this market.

R&D Investment & Innovation Signals

Leading companies are investing across thermal engineering, silicon co-design, and supply-chain localization rather than incremental product features:

●       Quanta Computer is targeting at least a doubling of AI server production capacity by end-2026 across Taiwan, Thailand, and US facilities, reflecting sustained hyperscaler capex rather than a short-term demand spike (official company guidance, November 2025).

●       Wiwynn is co-developing NVIDIA GB300 NVL72 systems with Wistron and showcased cold-plate liquid-cooling and two-phase immersion-cooling innovations at GTC 2025, positioning thermal engineering — not just assembly — as a core R&D investment area.

●       Supermicro's factory-integrated direct-to-chip liquid cooling enables GPU racks to run at 80–100 kW per rack, a density air-cooled OEM designs cannot match without a six-month retrofit disadvantage (per MRFR report page analysis).

●       AMD's internalization of ZT Systems' AI rack-scale design team signals silicon vendors are now investing directly in systems-level R&D rather than relying solely on ODM partners to translate reference architectures into deployable racks.

●       NVIDIA's USD 2 Billion strategic investment in Coherent Corp. (March 2026) targets silicon photonics and optical interconnects, addressing a bottleneck component shared across every major white-box ODM's next-generation rack architecture.

●       Foxconn, Quanta and Wistron are each expanding US-based manufacturing (Texas, Mexico) in direct partnership with Nvidia, reflecting R&D and capital allocation toward supply-chain resilience and tariff mitigation as much as toward product innovation.

●       The Open Compute Project's fourth-generation specifications — covering 48V power distribution and disaggregated BMC management — are being adopted by 60-plus data-center operators, and ODMs that certify fastest against these specs gain first-mover access to competitive hyperscaler bids (per MRFR report page).