The AI boom has reshaped the cloud infrastructure landscape as providers prioritize deploying servers optimized for AI model training. This shift has led to a predicted 30% increase in capital expenditure on cloud datacenter infrastructure this year. Vlad Galabov, head of Cloud and Datacenter Research at Omdia, believes the actual spending could exceed these forecasts.
Some cloud service providers have indicated they might raise their capex even further, potentially leading to unprecedented levels of investment. According to Omdia’s estimates, AI servers are expected to account for more than half of the overall server capex this year, making up 66% of total expenditure. However, these costly AI servers only represent 19% of shipments, reflecting their high price tag.
This surge in AI investment extends beyond the big three cloud providers—AWS, Azure, and Google Cloud—to include the top 10 cloud service providers and a growing number of specialized AI cloud service providers. The significant spending on AI has complex implications for cloud providers‘ ability to refresh their existing server fleets.
Cloud infrastructure spending shifts
Companies like Google have invested heavily in custom silicon to make their platforms more efficient. For example, Google developed a custom Video Encoding Unit (VCU) that can replace several standard servers, resulting in cost savings and consolidation. Despite the AI infrastructure boom, some enterprises are still hesitant to embrace public cloud resources fully and are opting for cloud repatriation—bringing workloads back in-house.
This trend is driven by the realization that public cloud resources can be as costly, if not more so, than running their own IT, often due to poor resource management. Galabov anticipates a greater focus on “cloud smart” strategies—optimizing workloads and placing them where it makes the most sense. In the near future, enterprises may increasingly rely on IT-as-a-service platforms, which offer the benefits of public cloud services in their own datacenters or colocation sites.
Second-tier cloud players such as Snowflake, MongoDB, and Oracle have shown the most substantial year-on-year growth rates, potentially challenging the dominance of the big three cloud operators. Specialized AI providers like CoreWeave may also contribute to this shift, introducing more competition to the market.
