Taiwan ODM/Brands AI server/PC supply chain: AI-server rack ramps and new smartphones are expected to lift Taiwan ODM/brand revenues through late 2026
Goldman Sachs forecasts average revenue growth for its 10 covered Taiwan supply-chain companies of 7%/2%/4% month-on-month in September/October/November 2026. The report favors AI-server exposure and selected smartphone supply-chain beneficiaries, while PC demand and component constraints remain uneven.
Summary
Goldman Sachs forecasts average revenue growth for its 10 covered Taiwan supply-chain companies of 7%/2%/4% month-on-month in September/October/November 2026. The report favors AI-server exposure and selected smartphone supply-chain beneficiaries, while PC demand and component constraints remain uneven.
- Average covered-company revenue rose 9% month-on-month in August.
- Average revenue growth is forecast at 49%/57%/52% year-on-year for September/October/November 2026.
- Wistron and Inventec receive higher earnings estimates after August results.
- AI-server rack specifications, higher ASPs and new smartphone launches are the principal growth drivers.
- High memory costs, PC demand pull-forward and component shortages remain key offsets.
Report Interpretation
Overview
This three-month revenue preview assesses 10 Taiwan ODM and brand companies in the AI-server, PC and consumer-electronics supply chain. Goldman Sachs expects late-2026 revenue growth to be led by next-generation GPU, ASIC and other rack-level AI servers, with smartphone launches and higher selling prices adding support, but sees divergent outcomes where PC demand, model transitions, valuation or component availability constrain the outlook.
Core views
Goldman Sachs reports that the 10 covered companies delivered average August revenue growth of 9% month-on-month. It expects average growth of 7%/2%/4% month-on-month in September, October and November 2026, and 49%/57%/52% year-on-year, respectively. The expected sequential expansion is driven by shipments of next-generation rack-level AI servers and flagship smartphones. The year-on-year growth also reflects continuing AI-server ramps, PC product-mix upgrades, new handset models and higher selling prices caused by elevated memory costs. The institution is positive on three supply-chain themes: Apple-related suppliers benefiting from new models and form-factor changes, including foldable phones in 2026 and iPhone 20 in 2027; AI-server components benefiting from specification upgrades; and rising penetration of ASIC AI servers. These themes underpin Buy ratings on Hon Hai, Wiwynn, Wistron and AVC. In contrast, the report retains Neutral ratings on Gigabyte, Quanta, Inventec, Compal and ASUS, and a Sell rating on Pegatron, reflecting company-specific valuation or operating constraints. For Hon Hai, Goldman Sachs expects monthly revenue to rise sequentially through December, producing 3Q26 and 4Q26 revenue growth of 14% and 23% quarter-on-quarter. Next-generation AI-server racks and CPO switches should carry higher ASPs, with Hon Hai expected to deliver the latest model around two to three months ahead of peers. GPU-powered racks are expected first, followed by LPU-powered racks from December onward. Smartphone launches and foldable phones are a second driver. The report projects 3Q26 revenue of NT$2.9 trillion, up 40% year-on-year and 14% quarter-on-quarter, and keeps 2026-28 earnings estimates largely unchanged. It maintains Buy and a NT$400 12-month target based on 17.9x 2027E P/E. Quanta faces a near-term PC and AI-server transition slowdown despite strong longer-term AI demand. Goldman Sachs expects 3Q26 and 4Q26 revenue growth of only 2% quarter-on-quarter after a high 1H26 base, while predicting a 54% month-on-month acceleration in December as next-generation higher-ASP AI racks ramp. July and August PC shipments fell 30% and 37% year-on-year. Quanta's capex is forecast at NT$44 billion in 2026E and NT$32 billion in 2027E, versus NT$15 billion in 2025, and its US factory lease illustrates capacity expansion for AI-server demand. The report forecasts aggregate capex by leading US cloud service providers to grow 76%/35%/8% year-on-year in 2026-28E, reaching US$1,145 billion by 2028E after estimate increases of 8%/27%/25%. It maintains Neutral and a NT$299 target, citing fair valuation. AVC is expected to benefit from AI-server rack adoption, increasing liquid-cooling penetration and new ASIC customer ramps. Goldman Sachs forecasts 3Q26 and 4Q26 revenue growth of 18% and 27% quarter-on-quarter and projects AI-server rack shipments, including Nvidia and AMD systems, of 55,000/105,000/163,000 in 2026-28E. It expects AVC's net income to compound at 39% annually in 2026-28E, supported by AI spending, its cooling-product breadth and leading position in total cooling solutions. The report maintains Buy with a NT$4,513 target based on 30.3x 2027E P/E. Wiwynn's August revenue exceeded Goldman Sachs estimates by 27%, supported by US cloud-service-provider general-server demand and a major customer's ASIC AI-server ramp. The report expects 3Q26 and 4Q26 revenue growth of 46% and 22% quarter-on-quarter, driven by new air- and liquid-cooled ASIC servers, resilient general-server demand related to agentic AI, GPU-powered rack production across Nvidia and AMD platforms in 4Q26, and potential new ASIC customers. It estimates NT$44 billion of 2026 capex, remaining high through 2028E, to diversify production across the US, Mexico and Malaysia and broaden AI-server platforms and customers. Goldman Sachs maintains Buy with a NT$3,777 target based on 21.5x 2027E P/E. Wistron's August revenue rose 166% year-on-year to NT$460 billion and was 48% above the institution's estimate, attributed to rack-level AI-server shipments. Goldman Sachs expects September-December sequential growth as GPU, CPU and LPU rack shipments ramp, while the company plans US$209 million of AI-server capacity investment in the US and Vietnam. The institution raises 2026-28 revenue forecasts by 11%/7%/6% and net-income forecasts by 2%/3%/4%, mainly for higher rack-server shipments and ASPs. Gross-margin forecasts decline because rack-level AI servers carry lower margins than other products. The target rises to NT$295 from NT$281 as the target multiple moves to 13.2x 2027E P/E from 13.1x; Buy is maintained. Gigabyte is expected to see AI-server racks and higher ASPs accelerate revenue growth, with 3Q26 and 4Q26 revenue growth forecast at 81% and 79% year-on-year versus 60% and 41% in 1Q26 and 2Q26. Goldman Sachs forecasts September and October revenue of NT$52 billion each, both up 87% year-on-year, and expects 2026 capex to rise 143% year-on-year to NT$2.8 billion. It retains Neutral with a NT$487 target because valuation is viewed as fair. For ASUS, the report forecasts 3Q26 revenue up 8% quarter-on-quarter but 4Q26 down 7%, as high memory costs brought PC demand forward and AI servers undergo a model transition. It nevertheless expects 3Q26 and 4Q26 revenue growth of 39% and 27% year-on-year from PC mix upgrades and AI-server ramping. Compal is expected to grow sequentially from September through December as AI-server shipments, capacity expansion and higher-end PC mix improve; Goldman Sachs forecasts 3Q26 and 4Q26 growth of 1% and 16% quarter-on-quarter. Its August revenue missed estimates by 20% amid lower PC shipments and weak consumer-electronics demand. Inventec's August revenue grew 36% year-on-year and was broadly in line with estimates, but declined 8% month-on-month because of shortages of memory, PCB and CPU components. Goldman Sachs expects a September sequential decline followed by October-December recovery, resulting in 3Q26 and 4Q26 revenue growth of 45% and 47% year-on-year. It raises 2026-28 revenue estimates by 8%/10%/6% and net-income estimates by 5%/4%/2%, while maintaining Neutral; its target rises to NT$61 from NT$56 as the target multiple rises to 15.3x 2027E P/E from 14.6x. Pegatron is expected to benefit from AI-server ramps, smartphone launches, broadband, IoT and game-console seasonality, with 3Q26 and 4Q26 revenue forecast to rise 18% and 25% year-on-year. However, its August revenue of NT$88 billion was 12% below Goldman Sachs estimates, and high memory costs continue to constrain PC and smartphone demand. Although the institution is positive on expansion into AI servers and automotive electronics, it maintains Sell on stretched valuation with a NT$76.9 target based on 10.8x 2027E P/E.
Analysis framework
Goldman Sachs compares reported July and August monthly revenue with its estimates, projects September-to-December revenue by company, and links the forecasts to AI-server platform ramps, ASPs, PC and smartphone demand, capacity spending and component supply. For covered stocks, it derives 12-month targets mainly from forward P/E multiples benchmarked against peer relationships between valuation, earnings growth and, where relevant, operating margin.
Methodology notes
Monthly revenue and shipment forecasting based on AI-server demand, PC and smartphone demand, component availability, pricing and capacity additions.
The report connects demand for AI servers and consumer devices with supply constraints, production capacity and ASP changes to estimate near-term company revenues.
AI-server demand is traced through rack manufacturers, cooling suppliers, ODMs, components and cloud-service-provider capex.
The report explains how cloud capex, server-platform transitions and specification upgrades transmit into shipments, revenue and capacity investment for Taiwan suppliers.
Forward P/E targets benchmarked against peers' earnings-growth and, in selected cases, operating-margin relationships.
Goldman Sachs applies a selected 2027E P/E multiple to forward EPS, using peer correlations between valuation and earnings growth to determine target prices.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- Hon Hai (2317.TW)AI-server racks, CPO switches and new smartphone models support growth.
- Strengths
- Expected early delivery of latest AI-server racks and exposure to foldable phones.
- Comparison
- Expected to deliver the latest model around two to three months ahead of peers.
- Risks
- Slower AI-server ramp, weaker EV total-solution performance, slower global capacity ramp or stronger EMS competition.
- Quanta (2382.TW)AI-server ramp and cloud capex support growth, offset by PC weakness and model transitions.
- Strengths
- High planned capex and expanding US capacity.
- Weaknesses
- PC shipment declines and muted 3Q26/4Q26 sequential growth.
- Comparison
- Neutral on fair valuation.
- Risks
- PC recovery, AI-server ramp and general-server demand may differ from expectations.
- AVC (3017.TW)Liquid-cooling adoption and new GPU and ASIC AI-server racks drive demand.
- Strengths
- Broad air- and liquid-cooling solutions and expected 39% 2026-28E net-income CAGR.
- Risks
- Slower liquid-cooling adoption, lower AI-server demand, competition or slower general-server recovery.
- Wiwynn (6669.TW)ASIC and GPU AI-server rack expansion supports revenue growth.
- Strengths
- August revenue beat, strong CSP demand and capacity diversification.
- Risks
- Weaker AI-server demand, slower general-server recovery or more intense server-ODM competition.
- Wistron (3231.TW)Rack-level AI-server shipments drive higher revenue and earnings forecasts.
- Strengths
- August revenue substantially exceeded estimates; US and Vietnam capacity investments support expansion.
- Weaknesses
- Higher rack-server mix reduces gross-margin forecasts.
- Comparison
- Target P/E remains within its historical average to average-plus-one-standard-deviation trading range.
- Risks
- Weaker PC recovery, slower AI-server ramp or weaker general-server demand.
- Gigabyte (2376.TW)AI-server rack growth, customer expansion and backlog visibility support 2H26 revenue.
- Strengths
- Rising capex and diversified production locations.
- Comparison
- Neutral on fair valuation.
- Risks
- PC recovery, gaming-PC growth and AI-server ramp may vary from expectations.
- Compal (2324.TW)AI-server ramp and move from motherboard to system integration support later growth.
- Strengths
- Business diversification and global capacity expansion.
- Weaknesses
- Consumer electronics remain the majority of revenue and are exposed to raw-material costs and weak demand.
- Comparison
- Neutral on fair valuation.
- Risks
- PC recovery, AI-server ramp and tablet demand may differ from expectations.
- Inventec (2356.TW)AI-server expansion supports revised forecasts, though supply shortages constrain near-term growth.
- Strengths
- Higher AI-server ASP assumptions support revised revenue and earnings.
- Weaknesses
- Short supply of memory, PCB and CPU components and sluggish PC demand.
- Comparison
- Target raised to NT$61 from NT$56 after earnings and valuation updates.
- Risks
- PC recovery, AI-server ramp and general-server demand may differ from expectations.
- Pegatron (4938.TW)AI-server and smartphone ramps support growth but do not offset valuation concerns.
- Strengths
- Expansion from consumer electronics into AI servers and automotive electronics.
- Weaknesses
- High memory costs cap PC and smartphone demand; August revenue missed estimates.
- Comparison
- Sell on stretched valuation.
- Risks
- Stronger smartphone recovery, faster AI-server ramp or stronger consumer-electronics demand.
Key data
- Average August revenue growth+9% MoMAcross the 10 covered companies.
- Average monthly revenue growth forecast+7% / +2% / +4% MoMSeptember/October/November 2026E.
- Average annual revenue growth forecast+49% / +57% / +52% YoYSeptember/October/November 2026E.
- Wistron August revenueNT$460bn, +166% YoY48% above Goldman Sachs estimates.
- Wistron 2026-28E revenue revisions+11% / +7% / +6%Driven by higher rack-level AI-server shipments and ASPs.
- Inventec 2026-28E net-income revisions+5% / +4% / +2%Mainly from higher revenue.
- Leading US CSP capex forecastUS$1,145bn by 2028EAggregate 2026-28E capex growth forecast at 76%/35%/8% YoY.
Impact & implications
The report sees AI-server rack ramps, higher specifications and ASPs, and growing ASIC-server and liquid-cooling adoption as the principal sources of late-2026 supply-chain growth. It differentiates among companies based on execution, capacity, PC and consumer-electronics exposure, component availability and valuation rather than treating the entire ODM group uniformly.
Risks
- High memory costs may continue to constrain PC and smartphone demand.
- AI-server shipment ramps or liquid-cooling adoption may be slower than expected.
- Component shortages, including memory, PCB and CPU supply constraints, may weigh on near-term production and revenue.
- Competition in server ODM, cooling and consumer-electronics manufacturing may intensify.
What to watch
- Monthly September-to-December revenue progression and the ramp of next-generation GPU, ASIC and other rack-level AI servers.
- PC shipment trends following first-half demand pull-forward and elevated memory costs.
- Cloud-service-provider capex and supplier capacity expansion.
- New flagship and foldable smartphone launches and their contribution to supplier demand.
- Availability of memory, PCB and CPU components.