Shipments of AI servers are bucking the trend across the tech industry with expected compounded double-digit growth through 2026. Research analyst firm TrendForce expects that by the end of 2026, about 200,000 AI servers will be shipped to CSP (cloud service providers), up from about 125,000 for 2022.
While Microsoft, Google, AWS and Meta account for the bulk of these purchases, ByteDance, the company behind the global phenomenon that is Tiktok, has emerged as a leader in the procurement of AI servers with a 6.2% market share, almost as much as the three other Chinese giants combined (Alibaba, Baidu, Tencent).
The inclusion of new features like the insanely popular “Bold Glamour” filter require a significant amount of processing, some of which is done on the smartphone itself and some on Tiktok’s own servers.
AI servers account for a tiny fraction of the overall server market (about 1%) but they usually cost hundreds of thousands of dollars each and carry high margins as well. They represent rare good news in an otherwise morose market. Global server shipments are only expected to grow by about 1% to just over 14 million units.
GPU vendor Nvidia, in particular, gets to enjoy the current AI boom (fuelled by ChatGPT and other similar AI technologies) with an estimated 80% market share as its technology is core to AI training.
An acceleration in AI adoption
Export control restrictions implemented by the US Commerce Department against Chinese companies are unlikely to put a damper on the appetite of China’s hyperscalers to buy AI servers at an accelerating rate. The A800, launched in November 2022 by Nvidia to meet the US Government’s clear test for reduced export control, has lower available interconnect bandwidth (400GBps vs 600GBps) compared to the A100, which may hamper China’s effort in the field of AI training but doesn’t stop it completely.
Anecdotal evidence points to a dumping of high performance AMD EPYC CPUs in the Chinese market, a sign perhaps that cloud service providers are investing massively in new servers capable of supporting a large number of GPGPU simultaneously.
TrendForce concludes its report saying “companies and organizations are scaling back IT spending as the global economy is being impacted by high inflation and sluggish growth. However, with applications such as chatbots and search engines driving the demand for an AI-based technological transformation, cloud companies will prioritize the related businesses or projects when allocating capital expenditure.”