Sugon/Nettrix Q3 Shipments Contract to 50k Units; High-End Pivot

ODM Direct Server Shipments Surpass 1.59 Million Units; Market S

Core Overview: Six Consecutive Quarters of Growth, Single-Quarter Shipments Break Records According to the latest data from DataTrack, global ODM Direct (white-box direct sales) server shipments reached 1,591.92 K pcs in the third quarter of 2025, setting the highest single-quarter record in history. This figure represents an 8.0% increase compared to the previous quarter's 1,474 K pcs, and a year-on-year increase of 19.7% compared to the same period last year (2024 Q3). Since the second quarter of 2024, shipments have shown positive growth for six consecutive quarters, indicating that under the wave of AI infrastructure construction, the procurement momentum for white-box servers is far superior to that of the traditional server market. Key Details: White-Box Market Share Nears 60%, Accelerated Computing Becomes Revenue Mainstay Latest reports from IDC and Dell'Oro Group indicate that in the third quarter of 2025, ODM Direct server revenue reached 66.8 billion USD, and global market share climbed to 59.4%, far leading Dell (8.3%) and Supermicro (4.0%). This wave of growth is primarily driven by servers with "Embedded GPUs" and "Non-x86 Architecture (such as Custom ASIC)." Notably, non-x86 server revenue saw a year-on-year growth rate as high as 192.7%, showing that Hyperscalers bypassing brand vendors to directly procure customized AI racks has become the absolute mainstream. In-Depth Attribution: Capex Race and the Blackwell Effect The core engine driving Q3 shipments to new highs comes from the aggressive Capital Expenditures (Capex) of players such as Meta, Microsoft, Google, and Amazon. Goldman Sachs points out that in order to prepare for the NVIDIA Blackwell Ultra platform and expand the deployment of in-house chips, the consensus value for 2025 capital expenditures by the four giants has been revised upward to historical highs. Dell'Oro analyst Baron Fung emphasized that AI-related spending has substantially crowded out some general-purpose server budgets, causing resources to be more concentrated on high-unit-price, high-density white-box AI models. Outlook and Risks: 2026 Focuses on Liquid Cooling and Inference Applications Short-term (1-2 months): With the traditional year-end peak season and the easing of AI chip supply, it is expected that Q4 shipments will remain at high levels, with an opportunity to challenge the single-quarter threshold of 1.65 million units. Medium-term (3-6 months): Looking ahead to 2026, TrendForce predicts that global AI server shipments will maintain a year-on-year growth rate of over 20%, and "Liquid Cooling technology" will become a new battleground, with penetration rates expected to approach 47%. However, investors need to pay attention to whether CSPs will accelerate the replacement of general-purpose GPUs with in-house ASICs due to cost considerations, as well as potential disruptions to the supply chain caused by geopolitical factors. Web Search Reference Sources https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQE8FTLFa__7KbDzmHQglkOEeH9g7CiHeY00v7B1cbwl9COeBsult72VCMzWaAdNmSQLe8drRJW2eqN5vRmud3mvpgffZ690KGuyO2kKnQYx72MPSfC-uovg6d5MTGtl8e2ogcZOwg3AGJWN3YUgthEt86k8uX5gxdlwWQ== https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQGK4tyOrTBzVBWmdBR45lcRyo2xDGxKhvJEWqPPwrH7rsvSeY0MxC6ZihDkgb99ygEdi5HHrrggEiMCBh5OCQH4EqDtE_YW7TjBi4--7cMaBgT0Yi8X8LZ0C_ZPcFijQ-umX8bGUprq2OWPooa_MQ42BLVh7GbhEfHdNI8fbEzexw67_STvmEATwXFJFKQgfA328qbsoI8ixfzHWv3UQ1FDgT6SeJK2RIF0DygdB8hYG7_oaJuJFVNNf5yzjNk01Cw= https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQF3kJvLnnszy98RxlzM_USbQSbDUgCVXXvVbxzJ2-qznhNyFpmibHElt2WZGD0i7exAIrAp7ZfHKdGfN8SrghV1XiCojC0YAeSn2tCL9YOcFXwgK9dkIoX8jYJijcBh8JhmuRixsgQemPPeuLv-O8-8QH6ZTfiRTdQqR0bDPETUwiEJq8lbkZHR3lRdfTqZguMowGz-k9A7qtfheHF5eVis4KgW