The AI infrastructure race is accelerating faster than most people realize.
According to McKinsey, global data center CapEx driven by AI could reach $5.2 trillion by 2030 in a baseline scenario.
Here’s how that capital is expected to be distributed:
• ~$3.3T in IT equipment (GPUs, servers, chips)
• ~$1.6T in data center infrastructure
• ~$300B in power generation
This buildout would require around 125 gigawatts of new capacity between 2025–2030 — roughly comparable to the output of ~125 nuclear reactors.
And that’s just the base case.
If demand accelerates:
→ Total investment could surge to $7.9T
→ Capacity expansion could reach 205GW
Even in a constrained scenario, we’re still looking at $3.7T in spending.
What’s driving this?
• Mass adoption of generative AI
• Enterprise-level AI integration
• Competition between tech giants and new entrants
• Government-backed AI infrastructure investments
This isn’t just hype — it’s a full-scale global infrastructure cycle.
The real question isn’t if AI will reshape industries…
It’s who captures the value from this $5T+ buildout.
