SOURCE RESEARCH
Artificial Intelligence and Machine Learning: Cloud AI
Market Data | 3Q 2025 | MD-AIMLCL-103
Get The ReportThe total revenue from chipsets dedicated to both AI inferencing and training workloads in the cloud is set to grow at a CAGR of 21%
throughout the rest of the decade. Revenue from silicon chips dedicated to AI workloads will reach over US$146 billion by 2031.
The majority of Capital Expenditure (CAPEX) originates from hyperscalers such as Google, AWS, Microsoft, Oracle, and Meta. These
investments cater to internal AI development and model training, as well as third-party customers serving inference and training
workloads. A growing proportion is from the neocloud segment, which is distorted by large sovereign AI deployments seen in the
Middle East, Asia-Pacific, and Europe. Notable examples include DataVolt in Saudi Arabia, Nscale in Norway and the United Kingdom,
and SK Telecom’s Haein cluster in South Korea.
