Global computing power is entering a new phase of rapid development

Benefiting from rapid economic recovery, the global server market continues to grow. In terms of complete systems, the total revenue of the global server market is expected to reach approximately USD 306 billion in 2024, representing a year-on-year growth of 73.5%. Revenue for the fourth quarter alone grew by 91% year-on-year, setting a new historical record.HPE/H3C leads the global server market with a 15.6% market share, followed by Dell (15.4%), Inspur (8.9%), Lenovo (6.4%), and Huawei (1.9%) in second to fifth place, respectively.

In terms of chipsets, the server chipset market has long been dominated by the x86 architecture, with Intel and AMD holding market shares of 81% and 16%, respectively. As Intel's dominance in the server space gradually weakens, AMD's market share is expected to continue rising. Additionally, ARM-based server chipsets are rapidly emerging, with global and domestic tech giants such as Nvidia, Amazon, Huawei, and Alibaba all launching their own ARM server CPUs.

The increasing scale of training data and model complexity is driving a surge in demand for AI servers. According to IDC, the global AI server market is projected to reach USD 30.74 billion in 2024 and grow to approximately USD 352.28 billion by 2034, with a compound annual growth rate (CAGR) of 27.62%. High-end AI server shipments are expected to rise from 639,000 units in 2024 to 1.323 million units in 2025—more than doubling—making them a major driver of growth in the overall AI market.

The growth of the AI market from 2020 to 2030

In 2024, the global AI server market saw ODMs, Dell Technologies, and Hewlett Packard Enterprise (HPE) occupy the top three positions with market shares of 44%, 20%, and 10% respectively. Together, these three vendors accounted for a combined 74% of the market.

On the chip side, traditional semiconductor giants are accelerating the development of their AI chip product portfolios, continuously expanding full-stack capabilities to seize dominance in the diversified computing ecosystem. Intel launched the next-generation high-performance deep learning training processor, Habana Gaudi2, which delivers twice the computing speed of its predecessor. After acquiring Xilinx, AMD plans to integrate Xilinx’s FPGA-based AI engines into its CPUs. NVIDIA introduced a new GPU architecture built on TSMC’s 4nm process, integrating 80 billion transistors, significantly boosting AI computing performance.

The era of exascale computing has arrived, and supercomputing equipment manufacturers are rapidly accelerating industrialization efforts—especially TSMC. In Q4 2024, TSMC reached a 67.1% market share in global semiconductor foundry services, ranking first.

According to the TOP500 list, Lenovo became the world’s leading supercomputer manufacturer with 161 systems, accounting for 32.2% of the global total. HPE ranked second with 96 systems (19.2%). Inspur, Atos, and Sugon followed with 50, 42, and 36 systems respectively, accounting for 10%, 8.4%, and 7.2%.

HPE launched the exascale supercomputer Frontier, its first E-class system following its $1.3 billion acquisition of Cray. French company Atos introduced the new BullSequana XH3000 supercomputer, which will support E-level traditional numerical simulations and 10E-level AI acceleration.

In terms of chips, Intel and AMD continue to dominate the CPU market. Among the TOP500 supercomputers, 388 use Intel CPUs (77.6%) and 93 use AMD processors. The use of heterogeneous computing chips in supercomputers is also increasing. Of the TOP500 systems, 168 feature accelerators or coprocessors, with 154 using NVIDIA chips and 8 using AMD chips.