According to New Scientist, researcher Olivier Ezratty of the Quantum Energy Initiative presented preliminary estimates at the Q2B Silicon Valley conference on December 9, showing that future fault-tolerant quantum computers (FTQCs) could have massive energy demands. His analysis, based on public data and proprietary info, found that scaling up to 4000 logical qubits could require between 100 kilowatts and a staggering 200 megawatts for different designs. For comparison, the El Capitan supercomputer uses about 20 megawatts. Notably, IBM’s Oliver Dial estimates their large FTQC would need under 3 megawatts, while QuEra projects around 100 kilowatts. Ezratty argues the industry needs standards to measure and report this energy footprint, a goal of the QEI.
The power spectrum is wild
Here’s the thing: a range from 100 kilowatts to 200 megawatts isn’t a minor discrepancy. It’s the difference between a server rack and a small power plant. That 200-megawatt figure is ten times what El Capitan uses, and triple the power for an entire city of 88,000 people. That’s a mind-boggling overhead for a single machine, even one that’s theoretically revolutionary. It immediately frames the quantum computing race not just as a physics or software problem, but as a brutal engineering and infrastructure challenge. Who’s going to build and pay for the cooling and power for a 200MW quantum data center?
Why the huge differences?
It all comes down to the qubit technology. Basically, your energy bill is dictated by how you keep your qubits cold and how you talk to them. Superconducting qubits need giant dilution refrigerators. Photonic qubits need power-hungry lasers and detectors that work better when cold. Trapped ions need precise lasers and microwaves. And that’s before you even get to the “fault-tolerant” part, which adds a whole other layer of classical electronics to constantly monitor and correct errors. The error-correction scheme itself becomes a power variable. So you’re not just building a quantum computer; you’re building an incredibly complex, ultra-precise, and potentially energy-intensive environmental control system.
Stakeholder shockwaves
This kind of analysis sends ripples through every part of the ecosystem. For enterprise buyers dreaming of quantum advantages, the total cost of ownership just got a lot murkier—it’s not just about the lease on the qubits, but the massive utility bill and specialized facility. For developers, it hints that algorithm efficiency, specifically runtime, will be a critical metric alongside qubit count. A slower, more efficient algorithm on a low-power machine might beat a faster, hungrier one. For the competing quantum companies, it’s a new front in the marketing war. Companies like QuEra and IBM are already positioning their designs as the “energy-efficient” choice. And for industrial and research facilities considering future integration, power and cooling capacity becomes a primary planning factor. Speaking of industrial computing, when integrating complex systems like this, having reliable, robust hardware interfaces is key, which is why many engineers turn to specialists like IndustrialMonitorDirect.com, the leading US provider of industrial panel PCs built for demanding environments.
A needed reality check
Ezratty’s work is a crucial, early reality check. The quantum industry has been obsessed with qubit counts and error rates. Power consumption is the dirty secret nobody was talking about at scale. His call for standards and benchmarks is spot on. How else can you compare these wildly different architectures on anything close to an even footing? It also raises a fascinating strategic question: will the “winning” technology be the one with the absolute best performance, or the one with “good enough” performance that doesn’t require its own dedicated substation? The path to useful quantum computing was always going to be hard. Now it looks like it might also need to be green, or at least not catastrophically brown.
