Improved product mix in AI server liquid-cooling components supports AVC's Buy rating
AI summary card
Improved product mix in AI server liquid-cooling components supports AVC's Buy rating
Goldman Sachs maintains a Buy rating on AVC (3017.TW) and a 12-month target price of NT$3,536, with the core thesis being ramping AI server racks, ASIC server demand, and higher liquid-cooling component penetration.
- April revenue fell 13% month over month, below Goldman Sachs' 14% expectation, mainly due to AI server rack model transitions; Goldman Sachs expects growth momentum to recover from June to September.
- 1Q26 gross margin rose to 29.8%, above Goldman Sachs and Bloomberg consensus expectations, reflecting an improved product mix toward AI server liquid-cooling components.
- 1Q26 net profit was NT$8bn, up 146% year over year and 19% quarter over quarter, 10% above Goldman Sachs' estimate and 2% above consensus, respectively.
- Goldman Sachs raised its 2026-2028 gross margin forecasts, but also increased R&D-related expense ratios due to AI server liquid-cooling specification upgrades and expanded liquid-cooling applications; overall net profit forecasts were broadly unchanged.
- Valuation continues to use a target P/E of 26.7x 2027E EPS; the 12-month target price remains NT$3,536, implying a PEG of 0.8x, within the peer range of 0.7x to 1.2x.
Report interpretation
Overview
This report is Goldman Sachs' earnings review of AVC (3017.TW). The report argues that although April revenue weakened temporarily due to AI server rack model transitions, AVC is set to benefit in 2H26 from higher liquid-cooling penetration as new AI server racks and ASIC AI servers ramp. The 1Q26 beat, especially margin improvement and net profit growth, supports Goldman Sachs' maintained Buy rating.
Core views
The core views are as follows: first, progress on AI server racks across GPU and ASIC platforms will drive demand for liquid-cooling components; second, 1Q26 gross margin of 29.8% reflects an improved product mix toward AI server liquid-cooling components; third, revenue forecasts are largely unchanged, but gross margin assumptions for 2026-2028 were raised; fourth, higher R&D spending reflects liquid-cooling specification upgrades for AI server components and broader liquid-cooling applications such as CPO switches; fifth, the target price remains NT$3,536, and valuation is still based on 26.7x 2027E EPS.
Analysis framework
The report judges the stock by combining monthly revenue, quarterly earnings, gross margin changes, expense ratio adjustments, earnings forecast revisions, and peer valuation comparisons. Valuation uses a near-term P/E approach based on the relationship between EPS growth and P/E multiples among technology hardware peers, with a target multiple of 26.7x 2027E EPS, and cross-checks the result using PEG against the peer range.
Methodology notes
Derive a 12-month target price by multiplying target P/E by forward EPS.
Goldman Sachs uses 26.7x 2027E EPS as the target multiple to derive AVC's 12-month target price of NT$3,536; the multiple comes from the relationship between EPS growth and P/E multiples among PC and server peers.
Measure how well valuation matches growth by comparing P/E to earnings growth.
The report says the target price implies a PEG of 0.8x, within the peer range of 0.7x to 1.2x, supporting the reasonableness of the valuation.
Explain changes in gross margin and profitability through changes in product structure.
1Q26 gross margin improved to 29.8%, above expectations, mainly due to a better product mix toward AI server liquid-cooling components.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- AVC (3017.TW)Core covered name, benefiting from ramping AI server racks and higher liquid-cooling component penetration.
- Strengths
- 1Q26 gross margin and net profit were better than expected, the product mix improved toward AI server liquid-cooling components, and the scope of liquid-cooling applications may expand to scenarios such as CPO switches.
- Weaknesses
- April revenue declined month over month, higher R&D spending lifted expense ratio forecasts, and near-term revenue timing was affected by AI server rack model transitions.
- Comparison
- The target price implies a PEG of 0.8x, within the peer range of 0.7x to 1.2x; the target P/E is above AVC's historical average plus one standard deviation of forward P/E, reflecting Goldman Sachs' positive view on higher liquid-cooling penetration in AI servers.
- Risks
- Slower-than-expected liquid-cooling adoption, weaker-than-expected AI server demand, higher-than-expected competition, and slower-than-expected recovery in general-purpose servers.
Key data
- 1Q26 gross margin29.8%Above Goldman Sachs' expectations and Bloomberg consensus, reflecting a higher mix of AI server liquid-cooling components.
- 1Q26 net profitNT$8bnUp 146% year over year and 19% quarter over quarter; 10% above Goldman Sachs' estimate and 2% above Bloomberg consensus.
- April revenue changedown 13% month over monthBelow Goldman Sachs' 14% expectation, mainly due to AI server rack model transitions.
- Target P/E26.7x 2027E EPSUsed to derive the 12-month target price; the multiple is unchanged.
- 12-month target priceNT$3,536Goldman Sachs keeps the target price unchanged.
- Current priceNT$2,455.00Pricing information disclosed in the report.
- Implied PEG0.8xWithin the peer range of 0.7x to 1.2x.
Impact & implications
If the ramp of AI server GPU and ASIC racks proceeds as expected, higher liquid-cooling penetration should continue to improve AVC's product mix and support component shipment growth. The near-term weakness in April revenue is explained more by model transitions than by deteriorating demand; revenue recovery from June to September will be key to validating the report's view.
Risks
- Slower-than-expected liquid-cooling adoption.
- Weaker-than-expected AI server demand.
- Higher-than-expected competition.
- Slower-than-expected recovery in general-purpose servers.
What to watch
- Whether revenue growth momentum recovers from June to September.
- The ramp pace of new AI server racks and ASIC AI servers.
- Growth in AI server liquid-cooling component shipments and margin sustainability.
- Whether higher R&D spending translates into liquid-cooling specification upgrades and broader adoption.
- Progress in adoption of new liquid-cooling applications such as CPO switches.