The global AI server market is expected to expand from US$142.88 billion in 2024 to US$837.83 billion by 2030, with a CAGR of 34.3%, according to MarketsandMarkets. This growth is driven by three key factors.
First, Nvidia's Blackwell GPU architecture drives enterprise adoption of AI technologies through significant performance gains, bolstering market growth. Second, the rising demand for generative AI (GenAI) has increased the need for high-performance infrastructure, fueling demand for AI servers in training and inference.
Third, investments in AI-optimized data centers and cloud services, driven by LLM training and inference growth, are prompting cloud service providers (CSPs) to expand AI hardware. Advances in cooling technologies, including air, liquid, and hybrid systems, are enhancing server performance and efficiency.
Taiwanese ODMs lead in custom AI server demand
DIGITIMES Research reports that in the third quarter of 2024, Meta Platforms, Amazon, Microsoft, and Google increased custom AI server orders, bringing Taiwanese ODMs' market share to 43.7%. This is expected to rise to 44% by the fourth quarter of 2024.
Inventec's key clients include HPE (38%), Dell (20.4%), Lenovo (43.8%), Google (42.3%), Microsoft (21.4%), and Amazon (38.6%).
Wiwynn's main clients are Meta (55%), Microsoft (12.4%), and Amazon (4.3%). Wistron serves HPE (21.8%), Dell (32.4%), IEIT Systems (Inspur; 4.5%), Lenovo (1.4%), and Super Micro Computer (Supermicro; 25.9%).
Foxconn's clients include HPE (37.8%), Dell (28.3%), Google (12.6%), Microsoft (35.1%), and Amazon (33.4%). Quanta serves Google (45.1%), Meta (45%), Microsoft (31.1%), and Amazon (19.6%). MiTAC primarily works with IEIT Systems (56.2%) and Amazon (3.3%).
In the fourth quarter of 2024, the top five AI server brands are projected to hold the following market shares: Dell (9.4%), Supermicro (8.7%), HPE (8.6%), IEIT Systems (6%), and Lenovo (4.4%).
Source: DIGITIMES Research, compiled by DIGITIMES, December 2024
Top 5 global AI server brands: 2024 key takeaways
I. Dell: advancing AI infrastructure amid cost management challenges
1. In March, Dell launched its AI Factory platform, integrating Nvidia's Blackwell architecture for LLM training and inference.
2. At Dell Tech World 2024 in May, Dell expanded its Nvidia partnership to streamline AI adoption for businesses, upgrading data centers and AI infrastructure.
3. In September, Dell announced continued layoffs and restructuring to cut costs, citing low AI server margins and weak PC demand.
4. In November, Dell debuted AI servers with Nvidia Blackwell GPUs and launched an AI Smart Solutions Center in Shenzhen.
5. In December, Dell COO Jeff Clarke confirmed deploying tens of thousands of GPUs for xAI's Colossus supercomputer in Memphis.
II. Supermicro: AI expansion and financial restructuring
1. Supermicro joined the S&P 500 on March 18, reflecting its strong Nvidia partnership.
2. At COMPUTEX 2024, Supermicro CEO Charles Liang sought additional Nvidia chips and announced a joint plan with ASE Group and Chunghwa System Integration (CSI) to build Taiwan's first liquid-cooled data center in Kaohsiung.
3. In August, Supermicro announced new supply chains in Taiwan and Malaysia. Later, reports of internal transactions and a delayed SEC filing raised financial concerns, including delisting risks.
4. In November, Nvidia reallocated Supermicro's clients to stabilize the AI supply chain, benefiting Gigabyte and ASRock. Supermicro also appointed BDO as auditor and extended its reporting deadline to 2025.
5. On December 8, Supermicro announced liquid cooling solutions for xAI's Colossus supercomputer in Memphis.
III. HPE: strategic acquisitions and pioneering AI infrastructure
1. In January, HPE announced a US$14 billion acquisition of Juniper Networks to strengthen its networking portfolio.
2. In March, HPE reportedly secured a multi-billion-dollar AI server order, outsourced to Foxconn.
3. In June, HPE announced a US$200 million AI server deal with Japan's AIST, integrating Nvidia H200 GPUs for GenAI. At HPE Discover 2024, it expanded its Nvidia partnership, debuting solutions like HPE Private Cloud AI.
4. In November, HPE launched the Cray Supercomputing EX system, the first 100% fanless direct liquid cooling (DLC) architecture.
IV. IEIT Systems (Inspur): pioneering AI servers and dominating the liquid-cooled market
1. In April, IEIT Systems launched the NF5180G7, a high-density server with 50°C immersion cooling, dual 64-core 350W Intel processors, and 32 DDR5 modules.
2. In May, Chairman Zhen Peng noted China's early-stage AI server market, with IEIT Systems' 14 factories capable of producing 2 million units annually at 50% higher efficiency if demand surges.
3. In August, 33 Chinese ISPs partnered with IEIT Systems, utilizing its YuanBrain EPAI platform to advance AI development.
4. In October, IDC reported China's liquid-cooled server market grew 98.3% to US$1.26 billion in early 2024, with IEIT Systems holding over 50% market share.
V. Lenovo: leading AI server production and cooling innovations
1. In March, Lenovo introduced ThinkSystem AI servers with Nvidia GPUs for GenAI and LLM development.
2. In September, Lenovo started AI server production in Pondicherry, India, and planned an AI lab in Bangalore.
3. In October, CEO Yuanqing Yang highlighted AI PCs and infrastructure as growth areas, focusing on servers and liquid cooling solutions.
4. At Tech World '24, Lenovo revealed its ThinkSystem N1380 Neptune cooling technology, allowing 1,000kW server racks to run without air conditioning.
Article translated by Levi Li
