Top 4 AI Predictions

Demand

Data center demand will exceed supply by over 500% by 2030

Hyperscaler annual capital expenditures are projected to reach $375 billion this year, a 36% increase from $275 billion in 2024. 2024 saw a 64% increase in spending compared to 2023. The top four hyperscalers alone are responsible for approximately 90% of this spend, and so far in H2 2025 these capex commitments are being built on rather than revised downwards.

Around half of this investment is being spent on infrastructure needs, such as self-build and leased data center capacity, and half on servers, including CPUs, TPUs and GPUs, for both cloud and AI related needs.

AI colocation revenue for training and inference is forecast to grow at an extraordinary 77% CAGR from 2025-2030, reaching $134 billion by 2030. By 2030, AI is expected to account for approximately 44% of total global data center colocation market revenue.

The steepness of the demand curve has huge implications for the data center industry. Demand for data center capacity is projected to outstrip supply from 2027-2030, with annual global demand reaching nearly 90 90 GW by 2030, exceeding supply by as much as 500%.

This imbalance is already evident. Colocation prices rose by an average of 35% between 2020 and 2023, and vacancy rates in Northern Virginia dropped to less than 1% in 2024. About half of all net new data centre capacity starting in 2025/26 is likely to be AI/ ML GPU clusters.

Typical training clusters are escalating in size, from approximately 40MW in 2020 to projected 1GW to 5GW systems before 2030. Overall, global digital infrastructure is expected to grow by a factor of 4 or 5 over the next 10 years, requiring a capacity increase of 130-150GW in the US alone.

Training versus Inference

There will be 4x more inference infrastructure than training infrastructure by 2030

This is another immense shift. The proportion of Inference to Training is switching as training models mature and the move to revenue generation (i.e. AI inference) takes off.

In terms of critical IT load capacity, AI inference is projected to grow at a 79% CAGR through 2030. This significantly outpaces AI training’s 25% CAGR over the same period. During 2026, inference capacity will overtake training capacity. By 2030, AI inference is expected to account for 80% of total AI critical IT load capacity, a reversal of the balance in 2023.

The shift of balance from training centers which are not so latency sensitive and can therefore go where power and space are plentiful, to inference centers, which need the lowest possible latencies, will put immense development pressure on existing densely populated “eyeball-rich” hubs, as well as accelerating the growth of new AI hubs in densely populated areas where data sovereignty is key and demand for services is high.