AI’s Infrastructure Problem Is Also Its Climate Problem
LabStart is working to address this challenge with our Data Center Sustainability cohort.
October 20, 2025
By Andrew Bray
As AI usage accelerates, so does the need for the infrastructure required to power it. The booming pace and scale of data center construction brings energy and environmental challenges that match its scale.
According to McKinsey, we’ll need up to 298 GW of data center capacity by 2030—nearly five times what exists today. That expansion could result in 40 million to 67 million tons of additional CO₂ emissions per year, driven largely by fossil-based electricity. At this pace, we risk blowing past the climate targets many tech companies have committed to.
The good news? We don’t have to choose between innovation and sustainability.
At LabStart, we back climate innovators working on hard-to-fund, high-impact solutions, and sustainable AI infrastructure is one of the most urgent challenges of this moment. That’s why we included a specialized Data Center Sustainability Cohort in our 2025 fellowship.
In a new piece I just published—The Opportunity: Building the Next Generation of Sustainable AI Infrastructure—I lay out five areas where we need breakthrough solutions to align AI infrastructure with climate goals:
- Repurposing heat to recover energy instead of wasting it
- Aligning operations with the electric grid to enable flexibility and carbon-aware computing
- Improving cooling to reduce water use, an overlooked stress point
- Siting data centers more strategically to incorporate climate, clean energy, and environmental impact factors
- Designing better compute to avoid more capacity by using hardware and software that can do more with less
We’re already seeing exciting innovation: Deep Green colocating with swimming pools to recover heat, Microsoft piloting 80%+ heat reuse, Elementl working with Oak Ridge National Laboratory and Google to site advanced nuclear near data-center demand, and others building flexible and grid-integrated compute.
This progress is only the beginning. We need more hardtech founders, grid innovators, and infrastructure builders to step up.
If you’re a scientist, engineer, or operator working at the intersection of AI and climate, now is your moment. Let’s build infrastructure that moves as fast as AI without sacrificing the planet in the process.
Thumbnail image created with AI.
Other Updates
.png)
.png)
.png)
.png)
.png)
.png)
.webp)
