Iron Mountain, in partnership with Structure Research has issued their top 4 predictions for the impact of AI on Global Data Infrastructure. The predictions look at what shifts companies can expect to see in the market over the next five years.
In the three years since ChatGPT launched, generative and agentic AI have become ubiquitous. Investment in GPUs and data centers is soaring, leading to significant changes in infrastructure and organizational adoption.
1. Data Center Demand Will Exceed Supply by Over 500% by 2030
Across the industry, hyperscaler capital expenditures are projected to reach $375 billion this year, a 36% increase from 2024. Half of this investment is spent on servers and GPUs, the other half is spent on data center capacity.This rapid growth will cause a massive supply deficit. Annual global demand will reach nearly 90 GW by 2030. This demand is expected to exceed available supply by as much as 500%.
2. There Will Be 4x More Inference Infrastructure Than Training Infrastructure by 2030
A major shift in the infrastructure needs is underway. Early AI investments focused on model training, but the market is evolving and is entering the “production phase.” The demand for real-time services will require massive inference deployment.
During 2026, inference capacity will officially overtake training capacity and by 2030 inference will account for 80% of all AI critical IT load. This is a complete reversal of the balance in 2023 and will mean Data Centers are needed closer to user-heavy hubs.
This move towards inference capacity means Data Centers will need to be built closer to end users, which means more will need to be built in densely populated areas such as cities.
3. Two-Gigawatt (2 GW+) Data Hubs Will Emerge in Every Global Region
To meet this demand, large data hubs will continue to scale rapidly and reach the following capacity by 2030:
North America: Northern Virginia will reach 8.5 GW. Dallas will scale to 2.8 GW, and Phoenix to 2.7 GW.
Europe: London (2.7 GW), Frankfurt (2.68 GW), and Paris (2 GW) will lead the market. Growth is also accelerating in Spain (Madrid and Barcelona), Germany (Berlin and Dusseldorf), and Portugal (Lisbon).
Asia-Pacific:, Tokyo (2.8 GW), Sydney (2.4 GW), and Johor (2.2 GW) will lead. Mumbai is projected to reach 2.15 GW.
4. The Cost of Artificial Intelligence Will Define Organizational Adoption
The cost of artificial intelligence is declining at an accelerated pace whilst the cost of the cheapest LLM has decreased by 10x every year. The declining price will not impact consumption, instead it will drive mass utilization and innovation.
The challenge here is the cost of AI tokens. Organizations are seeing overuse of AI impacting their bottom line. The move to usage-based pricing models mean leaders will need to keep an eye on how employees are using AI. Instead of using it for unnecessary tasks, guardrails will need to be put in place to ensure AI is used for the right tasks.