After Morgan Stanley, Goldman Sachs' Asia team also lowered its AI server shipment forecast for this year

After Morgan Stanley predicted that the cloud spending cycle may peak this year and significantly lowered its GB200 shipment forecast for this year, Goldman Sachs' Asia team also expressed their bearish view. Recently, the Goldman Sachs Asia team stated in a report that they have updated their estimates for the global server market size, with some AI server production plans being postponed from 2025 to 2026. This is mainly influenced by the iteration of GB200 and GB300 products, DeepSeek, and other factors, leading to a reallocation of different AI server budgets and adjustments in production bases. The Goldman Sachs Asia team expects (using 72-GPU equivalent calculations) that 31,000 rack-level AI servers will be delivered in 2025, a downward adjustment from the previous expectation of 38,000 units. In 2026, they expect to deliver 66,000 rack-level AI servers, an increase from the previous expectation of 58,000 units. Specifically looking at the performance in 2025, the aforementioned team expects the shipment volume of ODM rack-level AI servers to fluctuate: small deliveries in the first quarter, a ramp-up in the second quarter, product iterations in the third quarter, and another increase in the fourth quarter. In terms of the AI training server market, the team also expects production plans to be postponed to 2026 (using 8-GPU equivalent calculations): The shipment volume of AI training servers in 2025/26 is expected to grow by 42%/42% year-on-year, reaching 716,000 units/1.016 million units, respectively. AI training server revenue is expected to grow by 46%/39% year-on-year in 2025/26, reaching $179 billion/$248 billion, respectively. Among them, the value growth in 2025 is faster than the shipment volume, mainly due to the price increase brought by new product forms and GPU upgrades. Regarding the revenue outlook for the AI server market, the Goldman Sachs Asia team expects: AI training server revenue is expected to grow by 46%/39% year-on-year in 2025/26, reaching $179 billion/$248 billion, respectively. AI inference server revenue is expected to grow by 132%/14% year-on-year in 2025/26, reaching $18 billion/$21 billion, respectively. General server shipment volume is expected to grow by 5%/4% year-on-year in 2025/26, with the recovery in demand for this market mainly driven by the server replacement cycle, and its market value is expected to grow by 9%/6% during the same period. It is worth mentioning that the Goldman Sachs Asia team believes that with the application scenarios and implementation of AI, AI inference servers will see higher demand in 2025. The recently released low-cost DeepSeek R1 model further reduces the cost of large model training and inference, thus the shipment volume of AI inference servers is expected to grow by 58%/24% year-on-year in 2025/26, with the market size growing by 132%/14% year-on-year