Software, Not Concrete, Solves the AI Energy Crisis

Jan de Vries · 2026-07-02

The AI energy crisis isn't what you think it is. You've probably heard the doom and gloom: data centers are going to guzzle all the power, and the grid can't handle it. But here's the thing — that story is only half true. Stanford research shows that advanced economy power grids run at an average utilization of just 30%. Think about that for a second. Most of the time, we have plenty of electricity. We just can't get it where it needs to go. Duke University researchers found that electricity providers can already meet data center energy needs on 350 out of 365 days a year. The hardware is there. The electrons exist. What's missing is the intelligence to coordinate them — and that's a software problem, not a construction problem. ### The Build-It-All Approach Isn't Working The dominant response has been to build more. More generation. More substations. More transmission lines. Multinational giants like Amazon, Google, Meta, and Microsoft are signing power purchase agreements at a scale the energy industry has never seen. Data center electricity consumption is on track to more than double by 2030. But here's the catch: grid interconnection queues in Europe's main data center markets already run seven to ten years. In Dublin, where many major tech companies operate, the strain on the power grid has forced the operator to pause new connections until 2028. For companies unable to secure billion-dollar power deals, waiting for new infrastructure isn't a viable strategy. The capacity gap won't be closed by construction alone. It will be closed by software that makes existing infrastructure perform closer to its potential. And right now, almost nobody is funding that. ### The Case for Software After mapping more than 100 companies across the data center efficiency landscape, the conclusion is clear: software is the most capital-efficient solution to the AI power crisis, and it's the most underfunded layer in the stack. Consider this: the inference cost of running AI models declined 280 times between 2022 and 2024 — almost entirely through software improvements. Google doubled its data center energy efficiency through software optimization alone. The capital flowing into physical AI infrastructure dwarfs what's going into the software that makes that infrastructure perform. ### Four Layers, One Opportunity There are four distinct layers where software can recover capacity. Each one represents a massive opportunity. #### Grid Efficiency Grid efficiency software manages interconnection queues, data center flexibility, and distributed energy orchestration. This market is estimated to reach $4.12 billion by 2030. Much grid congestion is a coordination failure, not purely a physical one. Outdated systems can't route available power efficiently, and software can close much of that gap without new infrastructure. #### Facility Efficiency Facility efficiency covers the software and controls that reduce data center cooling and infrastructure energy use, lowering operating costs and increasing usable capacity. Public market estimates for adjacent facility-management software suggest a market of roughly $4 billion to $5 billion by 2030–2031. Crucially, every improvement in rack density creates additional sellable AI capacity without a new site, making facility efficiency a capacity expansion story, as well as a cost reduction one. #### Compute Efficiency Compute efficiency focuses on recovering stranded GPU capacity. Software that identifies unused capacity and orchestrates workloads more intelligently can lift GPU utilization materially. Research shows one production GPU-sharing system deployed across 20,000+ GPUs improved utilization from 26% to 76%. Meanwhile, according to Fujitsu, over 75% of organizations still report GPU utilization below 70% even at peak load. #### Software Efficiency Software efficiency — model compression, inference optimization, intelligent model routing — is one of the largest and least funded layers. The AI workload-management market implies a projected $197.5 billion opportunity by 2030, while the broader AI inference market is forecast to reach around $253.75 billion by 2030. ### The Bottom Line The broader AI infrastructure market is projected to reach between $223.5 billion and $394.5 billion by 2030. Yet the majority of AI infrastructure capital still flows into physical assets. The real opportunity — and the smartest investment — lies in the software that makes those assets work harder. We don't need to build a new grid. We need to make the one we have smarter. And that's a problem software can solve.