Tiktok has an insatiable appetite of AI servers research shows
Shipments from AI servers buck the trend across the technology industry with expected double-digit compound growth through 2026. Research analyst TrendForce expects about 200,000 AI servers to be shipped to CSP (cloud service providers) by the end of 2026, up from about 125,000 for 2022 .
Although Microsoft, Google, AWS and Meta account for the majority of these purchases, ByteDance, the company behind the global phenomenon Tiktok, has emerged as the leader in AI server purchases with a 6.2% market share, nearly equaling the three others Chinese giants together (Alibaba, Baidu, Tencent).
The inclusion of new features, such as the insanely popular “Bold Glamour” filter, requires 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 small portion of the total server market (about 1%), but they typically cost hundreds of thousands of dollars each and also have high margins. They represent rare good news in an otherwise bleak market. Worldwide server shipments are expected to grow only about 1% to just over 14 million units.
GPU Vendor Nvidia, in particular, can enjoy the current AI boom (fuelled by ChatGPT and other similar AI technologies) with an estimated market share of 80% as its technology is at the heart of AI training.
An acceleration in the adoption of AI
Export control restrictions put in place against Chinese companies by the US Department of Commerce are unlikely to dampen Chinese hyperscalers’ appetite to buy AI servers at an accelerated pace. Launched by Nvidia in November 2022 to meet the US government’s clear test for reduced export controls, the A800 has lower available interconnect bandwidth (400 GBps vs. 600 GBps) compared to the A100which may hamper, but not stop, China’s AI training efforts.
Anecdotal Evidence indicates a dumping of high performance AMD EPYC CPUs in the Chinese market, perhaps a sign that cloud service providers are investing massively in new servers that have a large number GPPU at the same time.
TrendForce concludes its report by saying, “Companies and organizations are scaling back their IT spending as the global economy is impacted by high inflation and slow growth. However, with applications such as chatbots and search engines driving demand for AI-based technology transformation, cloud companies will prioritize the related companies or projects when allocating capital expenditures.”