Gigabyte's AI Server Growth Accelerates Ahead of Schedule, Target Price Raised to NT$487, Maintain Neutral

Overview

This is a company research report by Goldman Sachs on Gigabyte (2376.TW), focusing on a more optimistic view of its AI server business and accordingly raising earnings forecasts and target price. Goldman Sachs believes Gigabyte's AI server business is ramping up faster than expected, with April revenue significantly exceeding prior estimates, leading to upward revisions of net profit forecasts and 12-month target prices (from NT$299 to NT$487). However, given that current stock price has already fully priced in these positives and valuation is at a high historical level, Goldman Sachs still maintains a neutral rating.

Core views

Business Progress: Goldman Sachs holds a positive view on Gigabyte's AI server business, expecting the company’s AI server revenue to grow at approximately 39% CAGR from 2025 to 2028, with AI server revenue share increasing from 54% in 2025 to 74% in 2028. The company is capable of developing AI servers across different architectures, with GB300 NVL72 entering mass production and Vera Rubin platform under development; additionally, it offers pre-integrated AI Pods (GIGAPOD) for high-density GPU cluster deployments, and a unified GPM platform for monitoring, scheduling, and resource management. This comprehensive solution helps customers deploy AI more efficiently. Revenue Beat: Gigabyte's April 2026 revenue rose 74% YoY and 34% QoQ to NT$52.3 billion, significantly exceeding Goldman Sachs' previous forecast by 51%, primarily driven by stronger-than-expected growth in HGX and GB platform AI servers. For Q2, Goldman Sachs expects revenue to rise 26% YoY (22% QoQ) to NT$129 billion; due to early order pull-forward in April, it anticipates revenue to decline sequentially in May and June (projected YoY growth of +5% and +8%, respectively, to NT$49 billion and NT$27 billion). Profit Revisions and Margins: Goldman Sachs raised net profit forecasts for 2026-28 by 36%/26%/29%, mainly reflecting three positive factors—company's capability to deliver AI servers across architectures and benefit from higher dollar content, capacity expansion supporting volume growth, and an integrated ecosystem represented by GIGAPOD enhancing cluster deployment efficiency. Gross margins for 2026 are revised upward due to product mix upgrading toward premium solutions and cost management, though gross margins for 2027/2028 are lowered by 0.2 percentage points each to reflect a growing proportion of lower-margin rack-level servers. Comparison with Market Expectations: Goldman Sachs’ 2026/27 earnings forecasts align generally with market consensus; 2026 revenue forecasts are slightly below market by about 3% (due to lower graphics card and motherboard revenues), whereas 2027 revenue forecasts are slightly above market by 1% (due to strong AI server revenue expectations). However, because rack-level AI server business has lower margins, gross margin forecasts are also relatively conservative.

Analysis framework

The core analytical framework of Goldman Sachs is 'AI server momentum + unit value increase + capacity expansion', using monthly revenue data as direct evidence of ramp-up progress: first confirming robust AI server growth through April's significant revenue beat, then revising full-year and next three-year revenue and earnings forecasts accordingly, finally arriving at valuation and target price. In terms of valuation methodology, Goldman Sachs adopts near-term P/E to derive the 12-month target price, setting the target P/E at 16.8x for 2027 (previously 12.8x). This multiple is not arbitrarily assigned but derived from the correlation between peer forward P/E and average EPS growth rate (the method remains unchanged), consistent with the historical trading range and average +1 standard deviation of ~16x P/E since 2010. Combined with revised earnings forecasts and target P/E, the target price is raised to NT$487. The report explicitly states that EPS growth is the main driver behind the stock’s performance.

Methodology notes

• Valuation MethodPE/PEG valuation

  Derive target price by multiplying forward P/E with earnings forecasts, and tie target P/E to EPS growth rate

  P/E reflects how much the market is willing to pay for each dollar of earnings. This approach ties target P/E to EPS growth, meaning higher growth leads to higher multiples. Goldman Sachs raised both earnings forecasts and target P/E from 12.8x to 16.8x, resulting in a significant increase in target price, illustrating a dual-driver logic of 'earnings × valuation multiple'.

• Industry/Industrial Analysis FrameworkVolume-price decomposition

  Decompose AI server revenue growth into two paths: unit value increase (dollar content upgrade) + volume expansion

  Revenue growth can be broken down into 'unit price (value per server/rack)' and 'volume (shipment quantity)'. This analysis suggests that AI server specifications upgrading will boost unit value, combined with expanded capacity driving shipment volume, both contributing to revenue growth—a key perspective for understanding its growth drivers.

• Quantitative/Factor/Portfolio Theory

  Peer regression of forward P/E vs. EPS growth (report fitted line y=39.472x+10.713)

  Goldman Sachs regresses peer companies’ forward P/E against earnings growth to derive a relationship where higher growth corresponds to higher reasonable P/E. By applying Gigabyte’s expected growth rate to this line, they determine its target P/E. This provides a benchmark-based pricing approach rather than arbitrary setting.

• Event Game Theory & Behavioral FinanceExpectation Gap/Expectation Management

  Use actual monthly revenue relative to prior institutional expectations to adjust judgment (April revenue 51% above estimate)

  Institutions compare actual company data with their own prior forecasts; larger gaps trigger adjustments in predictions and target prices. This report revised earnings and target price due to April revenue significantly exceeding expectations, highlighting a data-driven expectation correction logic.

Key data

• Rating / Target PriceNeutral / NT$487 (Previous NT$299)Maintain neutral, target price raised
• Current PriceNT$374Closing price on June 4, 2026
• April RevenueNT$52.3 BillionYoY +74%, QoQ +34%, 51% above Goldman Sachs' prior estimate
• 2Q26E RevenueNT$129 BillionExpected YoY +26%, QoQ +22%
• AI Server Revenue CAGR (2025-28E)Approx. 39%AI server revenue share rising from 54% in 2025 to 74% in 2028
• Net Profit Forecast Raise (2026-28E)+36% / +26% / +29%Due to revenue increase
• 2027E Target P/E16.8xPrevious 12.8x, consistent with historical average +1 SD (~16x)
• Revenue Forecast (2026E/2027E/2028E)NT$464 Billion / NT$579 Billion / NT$661 BillionYoY +38%/+25%/+14%
• EPS (2026E/2027E/2028E)NT$28.55 / NT$31.71 / NT$36.502026E YoY ~+57%

Impact & implications

Goldman Sachs believes that rapid growth in the AI server business will become the main driver of Gigabyte's revenue and profitability over the coming years, with AI server share continuously increasing altering the company's revenue structure. However, attention should be paid to structural impacts: low-margin rack-level AI servers will gradually increase in revenue share, which may depress overall gross margins in 2027/2028, showing a characteristic of 'high revenue growth but pressure on profit margins'. In terms of valuation, Goldman Sachs considers that the target price hike has already priced in most of the positive expectations, placing the valuation at a historically high level, thus maintaining a neutral rating despite favorable fundamentals.

Risks

• Faster or slower than expected PC market recovery
• Faster or slower than expected gaming PC growth
• Faster or slower than expected AI server ramp-up