850 Billion Into AI: Why Power Supply Is the Hidden Bottleneck in the AI Gold RUSH!!
- DPS
- Aug 19
- 3 min read
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 $235 billion in 2024 and will spend $337 billion in 2025 on AI hardware, software, and services. Prorating those numbers to this 18-month period gives us ~$416 billion in direct AI spending.
Infrastructure CAPEX (Build lens): Hyperscalers such as Amazon, Google, Microsoft, and Meta spent over $240 billion in 2024 and are guiding to $329–337 billion in 2025, with most of that dedicated to AI data centers and GPU clusters. That means $450+ billion has been invested in infrastructure alone over the past 18 months.
To put it simply: more than $400 billion has gone into AI technology, and another $450 billion+ into building out 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’s data center revenue alone has hit $154B over the last five quarters, validating the scale of GPU purchases.
CoreWeave, an independent AI cloud, has raised $25B+ and is targeting $20–23B capex in 2025.
Power & Grid Investment
JLL estimates $170B in data center construction financing is needed in 2025 alone.
Partnerships like BlackRock + Microsoft + MGX’s $100B platform are forming to solve the power and interconnection challenge.
The Bottleneck: Power Supply
While the GPU headlines grab attention, 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 focus on solving the energy bottleneck for hyperscale and AI infrastructure growth. With expertise in power systems for mission-critical environments, we provide:
Redundant Power Design
UPS systems and backup generation that ensure continuous uptime for AI workloads.
Natural Gas & Long-Term Contracts
Competitive 15–30 year natural gas delivery contracts that give data centers predictable, low-cost fuel supply.
Scalable Infrastructure
Solutions tailored to hyperscale and colocation data centers, ensuring that as clusters grow, the power foundation scales with them.
Trusted Partnerships
Authorized distributor relationships with leading manufacturers, ensuring the availability of critical equipment even in times of supply chain scarcity.
In a world where $450 billion is being spent to build AI infrastructure, power availability and reliability are what ultimately determine success. GPUs and servers can only function if the electricity supply is stable, redundant, and cost-effective. That’s exactly where Data Power Supply bridges the gap.
The past 18 months have seen unprecedented spending on AI and AI infrastructure:
~$416 billion on AI technology (hardware, software, services).
$450 billion+ on data center infrastructure buildouts, with power emerging as the limiting factor.
As the AI revolution accelerates, the winners will be the organizations that secure long-term, reliable, and redundant power solutions. Data Power Supply is committed to providing the energy backbone for the world’s next generation of data centers and AI workloads.
Comments