$850B has fueled AI in just 18 months—but the real choke point isn’t GPUs. It’s power. Discover the energy backbone of AI’s future.

How Much Has Been Spent on AI and AI Infrastructure in the Last 18 Months — And Why Power Is the Critical Bottleneck

The Unprecedented Scale of AI Spending

Over the last 18 months (Feb 2024–Aug 2025), the world has witnessed an explosion of investment into AI and the infrastructure that powers it:

• AI Technology Spend (IDC lens): IDC projects that organizations worldwide spent about $235B in 2024 and will spend $337B in 2025 on AI hardware, software, and services. Prorating those numbers to this 18-month period gives us ~$416B in direct AI spending.

• Infrastructure CAPEX (Build lens): Hyperscalers such as Amazon, Google, Microsoft, and Meta spent $240B+ in 2024 and are guiding to $329–337B in 2025, with most of that dedicated to AI data centers and GPU clusters. That means $450B+ has been invested in infrastructure alone over the past 18 months.

In short: More than $400B has gone into AI technology, and another $450B+ into the physical infrastructure that makes it run — in just a year and a half.

Where the Money Is Going

Hyperscaler Buildouts

• Amazon: ~$100B capex in 2025

• Google: ~$85B capex in 2025

• Meta: $66–72B in 2025

• Microsoft: ~$80B in FY2025, with single quarters hitting $24–30B

Suppliers & AI Clouds

• NVIDIA: $154B in data center revenue over the last five quarters, validating the scale of GPU purchases

• CoreWeave: Raised $25B+ and targeting $20–23B capex in 2025

Power & Grid Investment

• JLL: Estimates $170B in data center construction financing needed in 2025 alone

• BlackRock + Microsoft + MGX: Launching a $100B platform to solve the power and interconnection challenge

The Bottleneck: Power Supply

While GPUs grab headlines, the true constraint is power. AI data centers require hundreds of megawatts of stable, redundant power per campus, and growth is outpacing utility generation capacity.

Key challenges include:

• Securing long-term, low-cost electricity contracts

• Designing redundant, mission-critical power systems

• Integrating natural gas, renewables, and storage for resilience

• Avoiding downtime and protecting against grid instability

Data Power Supply: Delivering the Energy AI Runs On

At Data Power Supply, we solve the energy bottleneck for hyperscale and AI infrastructure growth. Our expertise includes:

• Redundant Power Design: UPS systems and backup generation to ensure continuous uptime for AI workloads

• Natural Gas & Long-Term Contracts: Competitive 15–30 year delivery contracts providing predictable, low-cost fuel supply

• Scalable Infrastructure: Tailored solutions for hyperscale and colocation data centers that scale with cluster growth

• Trusted Partnerships: Authorized distributor relationships ensuring critical equipment availability even during supply chain scarcity

Conclusion

In the past 18 months, spending has reached unprecedented levels:

• ~$416B on AI technology (hardware, software, services)

• $450B+ on infrastructure buildouts — with power as the limiting factor

As the AI revolution accelerates, the winners will be those who secure long-term, reliable, and redundant power solutions. Data Power Supply is committed to being the energy backbone for the world’s next generation of data centers and AI workloads.