Global AI server shipments expected to jump 36%
Global AI server shipments are projected to increase by 36 percent next year to 3.83 million units, driven by substantial investment from nine major cloud service providers in China and the US, according to TrendForce Corp. This growth comes as Nvidia reportedly instructs data center partners to limit sales to Chinese customers, and MediaTek secures a new AI accelerator customer outside China.
Despite market concerns, nine major cloud service providers, including US-based Google, Amazon, Meta, Microsoft, Oracle, and Chinese firms like ByteDance, Tencent, Alibaba, and Baidu, are expected to boost their combined capital expenditure by 60 percent to US$1.52 trillion next year, TrendForce research manager Frank Kung stated. This investment is anticipated to fuel a 92 percent growth in the AI server market, reaching US$919 billion and comprising 81 percent of the total server market value.
The supply chain for AI infrastructure is experiencing increased procurement and rising prices across components like memory chips and GPUs, Kung added. Nvidia is forecast to maintain its dominant position in AI server supply, with its Vera Rubin racks expected to account for about 70 percent of its total rack shipments next year, potentially surpassing the GB series in the second quarter. However, shipments of the next-generation Rubin Ultra platform are expected to be limited.
Amidst this growth, geopolitical factors appear to be influencing market dynamics. The Financial Times reported that Nvidia is directing its data center partners to scale back work with Chinese customers, requiring cloud computing clients to sell at least 51 percent of their services to non-Chinese entities for future contracts. Concurrently, MediaTek Inc. announced a new customer for its customized AI accelerator for data centers, explicitly noting this customer is not from China, which contrasts with earlier speculation linking ByteDance to the deal.
Share this article
Related reading
6 stories
OpenAI’s Dots are coming up against the price of privacy, and the price of compute
We previously explored the rising compute costs and privacy concerns impacting AI infrastructure, a key theme here.

Google launches Gemini Omni 1.2 Flash with longer clips and lower cost
Google, a major player in this AI server investment surge, recently launched its Gemini Omni 1.2 Flash model.

AI bubble bursts, fifth plenum resolution, Singles’ Day

Hong Kong’s Exchange Fund earnings tumble 74% as HKMA warns of more market turbulence

