Goldman Sachs is positive on AI server and new-model drivers for Taiwan ODM/brand names, but flags divergent MoM momentum in July-August

Overview

This is a Goldman Sachs company research report providing a 3-month revenue outlook for seven Taiwan ODM/brand companies. The report focuses on monthly revenue trends from June to August 2026 and believes that June was broadly flat on average MoM but still strong YoY, with AI servers, ASIC AI servers, liquid cooling components, PC mix upgrades, and smartphone new models as the key drivers. Covered companies include Hon Hai, Quanta, Wiwynn, AVC, ASUS, Compal, and Inventec.

Core views

Goldman Sachs believes continued AI server ramp-up remains the most important source of growth for Taiwan's ODM/brand supply chain, especially Hon Hai's AI server racks, Wiwynn's ASIC AI servers and U.S. CSP demand, and AVC's rising liquid cooling component penetration. Although the PC and notebook supply chain is affected by high memory costs and front-loaded demand in the first half, AI PCs, gaming PCs, and product mix upgrades can still support YoY growth for some companies. Smartphone new models, including future form-factor change models, are also seen as important support for Apple supply chain demand.

Analysis framework

The report centers on company monthly revenue, company data, Goldman Sachs estimates, and YoY/MoM trends, comparing revenue changes from March to August 2026, while also incorporating quarterly revenue, AI server model transitions, PC shipments, liquid cooling penetration, U.S. CSP demand, and high-base effects to assess near-term growth momentum.

Methodology notes

• Revenue forecasting framework3-month revenue outlook

  Monthly revenue YoY and MoM

  The report identifies the impact of AI server volume ramp-up, model transitions, front-loaded PC demand, and seasonal high-base effects on each company by comparing the monthly revenue YoY and MoM changes in Jun/Jul/Aug 2026E.

• Valuation methodologyP/E and PEG&M peer comparison

  12-month target price

  Target prices for different companies are mainly based on 2026E or 2027E P/E multiples, with reference to peer EPS growth, P/E correlation, margins, or the PEG&M framework; Hon Hai's target price uses 21.0x 2026E P/E, AVC uses 26.7x 2027E P/E, Wiwynn and ASUS use 15.8x 2027E P/E, and Compal and Inventec use 13.6x 2027E P/E.

• Investment attribute frameworkGS Factor Profile

  Growth, financial returns, valuation multiples, and composite percentile

  The Goldman Sachs factor profile compares individual stocks with the market and industry peers across growth, financial returns, valuation multiples, and composite indicators to provide investment context, but the core conclusions of this report still come from the company revenue outlook and valuation methodology.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Hon Hai (2317.TW)

  Beneficiary of AI server racks and consumer electronics EMS; rated Buy with a target price of NT$400.

  Strengths

  AI server rack demand is solid, with 2Q/3Q26E revenue YoY expected at +31%/+44%, and May revenue was 5% above Goldman Sachs' prior estimate.

  Weaknesses

  June is expected to be -23% MoM due to a high base, while comprehensive EV solutions and global capacity ramp-up still require execution.

  Comparison

  Compared with PC-oriented ODMs, Hon Hai benefits more directly from AI server rack volume growth.

  Risks

  AI server business ramp-up slower than expected, EV business underperforming expectations, slower-than-expected global capacity expansion, and intensifying competition in consumer electronics EMS.

• Wiwynn (6669.TW)

  Beneficiary of ASIC AI servers and U.S. CSP server demand; rated Buy with a target price of NT$7,147.

  Strengths

  Leadership in ASIC AI servers, diversified GPU AI server rack exposure, and U.S. CSP general-purpose server demand support QoQ growth in 3Q26E.

  Weaknesses

  Some projects shifting to a consignment model lower reported revenue, and May revenue was 12% below Goldman Sachs' estimate.

  Comparison

  It has higher exposure to ASIC AI servers among AI server ODMs, but changes in revenue recognition make short-term revenue comparability lower.

  Risks

  Weaker-than-expected AI server demand growth, slower-than-expected recovery in general-purpose servers, and intensifying competition among server ODMs.

• AVC (3017.TW)

  Beneficiary of AI server liquid cooling components and rising liquid cooling penetration in ASIC AI servers; rated Buy with a target price of NT$3,536.

  Strengths

  Jun-Sep is expected to post consecutive MoM growth, benefiting from new AI server rack liquid cooling components, rising liquid cooling penetration in ASIC AI servers, and new project ramp-up.

  Weaknesses

  The business is highly sensitive to the pace of liquid cooling adoption and AI server demand.

  Comparison

  Compared with full-system ODMs, AVC is more exposed to the high-elasticity AI server liquid cooling component segment.

  Risks

  Slower-than-expected liquid cooling adoption, lower-than-expected AI server demand, intensifying competition, and slower-than-expected recovery in general-purpose servers.

• Quanta (2382.TW)

  AI server and PC supply chain company; rated Neutral with a target price of NT$299.

  Strengths

  JunE is expected at +98% YoY/+21% MoM, with a low base and quarter-end improvement in PC consumption providing a short-term rebound.

  Weaknesses

  May revenue was 11% below Goldman Sachs' estimate, and 3Q26E is affected by AI server rack model transitions and slower demand after front-loaded PC shipments.

  Comparison

  YoY growth is very strong, but short-term MoM volatility and the 3Q26E revenue pullback keep the rating at Neutral.

  Risks

  Changes in the strength of PC market recovery, changes in the pace of AI server ramp-up, and changes in general-purpose server demand.

• ASUS (2357.TW)

  PC brand, beneficiary of AI PCs and rack-level AI servers; rated Neutral with a target price of NT$1,082.

  Strengths

  AI PC mix upgrades and rack-level AI server ramp-up support YoY growth in 2Q/3Q26E.

  Weaknesses

  Pressure from high memory costs remains, and May fell 16% MoM due to the high base in March-April.

  Comparison

  As a brand vendor, it benefits from mix upgrades, but compared with AI server components and server ODMs, it has greater exposure to the PC cycle.

  Risks

  PC market growth below expectations, AI and gaming PC growth below expectations, and slower-than-expected AI server ramp-up.

• Compal (2324.TW)

  PC ODM and AI server capacity expansion company; rated Neutral with a target price of NT$39.5.

  Strengths

  May PC shipments rose MoM to 2 million units, and revenue mix diversification plus expected startup of the new U.S. factory support growth in 2Q/3Q26E.

  Weaknesses

  Jul-Aug is expected to decline slightly MoM after the high base in June, and the company remains sensitive to PC demand.

  Comparison

  Growth momentum has improved versus traditional PC ODMs, but AI server exposure and earnings elasticity are weaker than those of the Buy-rated names.

  Risks

  Weaker-than-expected PC market recovery, slower-than-expected AI server ramp-up, and weaker-than-expected tablet demand.

• Inventec (2356.TW)

  AI and general-purpose server capacity expansion company; rated Neutral with a target price of NT$56.0.

  Strengths

  May revenue rose +35% YoY, and U.S. AI server capacity ramp-up supports continued YoY growth in Jun-Aug.

  Weaknesses

  Jun-Aug is expected to decline MoM consecutively, affected by front-loaded PC demand and rack-level AI server model transitions.

  Comparison

  Server business expansion provides medium-term support, but the short-term MoM trend is weaker than that of the Buy-rated names.

  Risks

  Weaker-than-expected PC market recovery, slower-than-expected AI server ramp-up, and weaker-than-expected demand for general-purpose servers.

Key data

• Average revenue MoM for the seven companies+6% / -5% / +0%Corresponding to Jun/Jul/Aug 2026E; reflects improvement in June, pullback in July, and average flatness in August.
• Average revenue YoY for the seven companies+39% / +37% / +29%Corresponding to Jun/Jul/Aug 2026E; strong YoY growth is mainly driven by AI servers, PC mix upgrades, smartphone new models, and higher ASPs due to elevated memory costs.
• Hon HaiMay +40% YoY/+3% MoM; JunE +22% YoY/-23% MoM; 2Q26E/3Q26E +31%/+44% YoY; Buy, TP NT$400Demand for AI server racks continues to support growth during the traditional electronics off-season, but June is expected to decline MoM due to a high base.
• QuantaJunE +98% YoY/+21% MoM; 3Q26E -20% QoQ; Neutral, TP NT$299A low base drives the rebound in June, but 3Q26E is affected by AI server rack model transitions and slower PC shipments.
• AVCJunE +75% YoY/+17% MoM; 3Q26E +56% YoY/+21% QoQ; Buy, TP NT$3,536New AI server rack liquid cooling components, rising liquid cooling penetration in ASIC AI servers, and ramp-up of new projects are the core drivers.
• WiwynnJunE +6% YoY/+8% MoM; 3Q26E +23% QoQ; Buy, TP NT$7,147Some server projects shifting to a consignment model reduce reported revenue, but gross margin and long-term working capital pressure are expected to improve; ASIC AI servers and U.S. CSP demand support growth.
• ASUSJunE +24% YoY/+23% MoM; 2Q26E/3Q26E +26%/+12% YoY; Neutral, TP NT$1,082AI PC mix upgrades and rack-level AI server ramp-up support growth, but May declined MoM due to the high base in March-April.
• CompalJunE +23% YoY/+5% MoM; 2Q26E/3Q26E +20%/+32% YoY; Neutral, TP NT$39.5PC shipments rose MoM to 2 million units in May, and the new U.S. factory is expected to start production by the end of 2Q26, supporting AI servers and revenue mix diversification.
• InventecMay +35% YoY; JunE +22% YoY/-7% MoM; 2Q26E/3Q26E +31%/+10% YoY; Neutral, TP NT$56.0AI and general-purpose server demand remains strong, but revenue is expected to decline MoM from June to August due to front-loaded PC demand and rack-level AI server model transitions.
• Taiwan ODM notebook shipmentsMay-26 was 9.3 million units, Apr-26 was 9.1 million units, Mar-26 was 13.8 million unitsThe table shows total shipments improved slightly MoM in May, but remained below the March peak.

Impact & implications

For portfolios, the report reinforces the view that the AI server chain is superior to the traditional PC chain: companies with exposure to AI server racks, ASIC AI servers, liquid cooling components, or U.S. CSP demand benefit more; companies more dependent on PC shipments or facing model transitions or high-base pressure have less stable near-term momentum. Buy ratings are concentrated in Hon Hai, Wiwynn, and AVC, where certainty around AI servers and liquid cooling is higher.

Risks

• AI server business ramp-up slower than expected could weaken growth for Hon Hai, Wiwynn, Quanta, ASUS, Compal, and Inventec.
• ASIC AI server penetration or the pace of liquid cooling adoption slower than expected could affect the high-growth assumptions for Wiwynn and AVC.
• PC market recovery weaker than expected, or front-loaded demand in 1H26 pulling forward later demand, could weigh on Quanta, ASUS, Compal, and Inventec.
• High memory costs could suppress PC supply chain shipments, while also potentially lifting ASPs, affecting judgments on revenue and profit structure.
• Intensifying competition among server ODMs, intensifying competition in consumer electronics EMS, or customer project shifts could compress profitability.
• If global capacity ramp-up, U.S. factory startup, or EV business execution fall short of expectations, companies' medium-term growth delivery could be affected.

What to watch

• Whether actual monthly revenue in Jun/Jul/Aug 2026 matches Goldman Sachs' average MoM estimates of +6%/-5%/+0% for the seven companies.
• Whether AI server rack and ASIC AI server projects navigate the model transition period smoothly.
• The strength of U.S. CSP orders for general-purpose servers and AI servers.
• Changes in AVC's liquid cooling component penetration in new AI server racks and ASIC AI servers.
• Whether PC and notebook shipments see a more pronounced pullback after front-loaded demand in 1H26.
• The demand boost from Apple supply chain new models and future form-factor change models.