AI capacity crisis: 50% of planned data centers delayed – energy is the bottleneck

A new report reveals that up to half of announced data center projects may be delayed, with energy access emerging as the critical bottleneck for AI growth.

Key figures from the Sightline Climate report

Analysis firm Sightline Climate is tracking 190 gigawatts (GW) of planned data center projects globally. The findings are alarming:

• Only 5 GW is under construction out of 190 GW planned capacity
• 6 GW came online in 2025 – a solid year, but far from meeting demand
• 36% of projects experienced delays in 2025
• 50% of announced projects may be delayed
• AI is expected to drive up data center power consumption by 175% by 2030 (Goldman Sachs)

The main culprit is not a lack of capital or technology – it's access to power.

What are the tech giants doing?

Google, Meta, and Microsoft are responding by building their own energy infrastructure:

• Google invested $1 billion in Form Energy's 100-hour battery
• Meta is allocating $115–135 billion to AI infrastructure in 2026
• Microsoft is purchasing capacity from nuclear, wind, and solar sources

Startups like Amperesand, DG Matrix, and Heron Power are developing new power conversion technologies, while Camus, GridBeyond, and Texture are building software for smart energy grids.

What does this mean for enterprise leaders?

For CIOs and AI decision-makers, this is a critical signal:

• Cloud capacity may tighten – AWS, Azure, and GCP will face capacity constraints in growth markets
• Prices may increase – data center scarcity drives up cloud pricing
• Nordic countries are advantaged – clean energy and cool climates make Scandinavia attractive for data center development
• Investment opportunities – energy tech is the next major AI-adjacent investment category

The bigger trend

We are seeing a marked shift: VC money that previously flowed to AI models and applications is now moving toward energy infrastructure. AI is becoming an energy problem as much as an algorithm problem.