Compute First, Climate Second
Corporate energy strategies are bending the grid and breaking decarbonization timelines
Data Center Dynamics advertised a late 2024 survey by Aggreko revealingly titled “Rebalancing the Energy Transition.” It reads like an attempt to justify delaying net-zero targets while AI-driven data centers push global power demand to unprecedented levels. Based on a survey of 400 CEOs from high-energy industries, including data centers, the report aims to gauge how industry leaders are adapting to rising energy costs, supply chain bottlenecks, and regulatory scrutiny. But in the face of AI’s breakneck expansion, the report reveals less about how companies are transitioning toward sustainability and more about how they are maneuvering to protect profit margins while demanding ever-greater energy supplies.
The inclusion of data centers in this analysis is critical, though its treatment of them is remarkably understated. AI-driven infrastructure is now the driving force behind global energy expansion, with Goldman Sachs projecting a 165% increase in data center power demand by 2030 and RAND estimating that AI data centers alone could require 327 GW by 2030—nearly quadrupling total global data center capacity from 2022. The report acknowledges that businesses are shifting their energy strategies but fails to grapple with the sheer scale of AI’s acceleration and its fundamental incompatibility with any meaningful energy transition.
Net-Zero, Unless It Costs Too Much
The first major admission comes quickly: “95% [of CEOs] have changed their timescales for net zero in light of recent energy supply and pricing issues.” There it is. The classic move—net-zero as a flexible aspiration rather than a commitment. And yet, this should come as no surprise. As AI infrastructure expands, demand for energy is surging far beyond what the grid can handle, forcing companies into a predictable dilemma: either slow down AI growth (unthinkable in a winner-takes-all competition) or quietly revise sustainability targets to accommodate the next phase of data center mania.
Even when the report tries to reassure us that “80% expect to increase investment into the energy transition over the next 12 months”, it immediately hedges: “most of those will only increase budgets marginally.” That’s the trick—make just enough of an effort to maintain ESG credibility while ensuring that energy-intensive industries, particularly AI, continue their unchecked expansion. The fact that AI is pushing power demand beyond grid capacity doesn’t lead to a reevaluation of priorities—it leads to an industry-wide effort to bend sustainability timelines around the problem rather than addressing it head-on.
Power First, Climate Second
If there were any doubt about where corporate priorities lie, the report makes it explicit: “Cost and commerciality remain higher priorities than decarbonisation overall.” This isn’t just a business decision—it’s an existential one. Data centers, already consuming over 55 GW globally, are projected to hit 84 GW by 2027. AI’s power demand is expanding so fast that even the most aggressive renewable energy strategies will struggle to keep up, leading to a sharp resurgence of fossil fuel infrastructure. The U.S. alone is planning to add 46 GW of new natural gas capacity by 2030, a move that directly contradicts the net-zero rhetoric plastered across corporate sustainability reports.
And what about supply chains? The report notes that “almost half said the supply chain was among the greatest risks to their energy transition.” But this framing conveniently ignores that AI’s energy supply chains aren’t just facing temporary logistical problems—they are structurally unsustainable. Transmission delays, slow grid interconnections, and local resistance to new energy projects are fundamentally at odds with the hyperspeed expansion of AI infrastructure. AI companies can demand power all they want, but they are running into the simple reality that energy infrastructure cannot be scaled as rapidly as GPUs can be installed.
The Nuclear Pivot
Faced with an energy crunch of its own making, Big Tech is already pivoting toward nuclear power as its long-term solution. Microsoft has openly signaled interest in small modular reactors, while OpenAI and hyperscalers are exploring direct partnerships with nuclear energy startups. As AI’s power demands surge past the limits of solar and wind, nuclear energy is increasingly being framed as the only viable pathway to sustaining data center expansion.
But this shift comes with risks. The political and economic barriers to scaling nuclear energy remain substantial, and the industry’s track record of project delays and cost overruns raises serious doubts about whether nuclear can arrive in time to prevent a deepening energy crisis. Worse yet, the growing reliance on nuclear as a corporate energy solution risks placing AI’s future in the hands of a few powerful players who control both compute capacity and energy production—entrenching even greater monopolization in an industry already defined by extreme concentration of power. While many communities are already reluctant to site transmission lines, solar, wind, and even battery systems in their backyard for cloud and AI, how would communities react to a nuclear reactor?
A Future Gridlocked by AI Energy Demand
Aggreko’s report, though framed as a guide to energy transition, ultimately confirms what climate political economists have long suspected: the more AI expands, the further net-zero targets slip away. The sheer scale of AI’s projected power demand—a possible 327 GW by 2030—is completely misaligned with the reality of today’s energy grid. Even if AI companies commit to renewables, the timeline for deploying enough wind, solar, and nuclear power to meet these demands does not exist within the current energy transition framework.
Rather than confronting these contradictions, corporate leaders have opted for a strategy of managed delay—expanding data centers first and figuring out the energy problem later. Yet, the grid is already buckling under existing power loads and the global energy transition is way behind schedule. The more AI demands, the more fossil fuels will be brought back online to meet that demand. I will cover some of this context more in my next article on the Trump administration’s AI industrial strategy.
The AI boom isn’t just an economic race—it’s an energy arms race (also a regular arms race). And right now, the companies leading this expansion aren’t transitioning to net-zero. They’re postponing it indefinitely, betting that energy production can somehow keep pace with AI’s exponential hunger.
Key References:
Aggreko. 2025. “Rebalancing the Energy Transition: What Does the C-Suite Need to Balance Sustainability and Profitability?” https://media.datacenterdynamics.com/media/documents/FINAL_Aggreko_Rebalancing_the_Energy_Transition_November_2024_-_Matthew_Westwood.pdf.