Power Consumption: The Blind Spot of Quantum Computing and How Aqsolotl Addresses It

When people talk about the future of quantum computing, the conversation usually centers on qubit coherence, error correction, or the race toward logical processors. Power consumption is rarely mentioned. Yet it is fast becoming one of the most concrete — and most underestimated — challenges the industry will have to face. 

At Aqsolotl, we manufacture quantum controllers with a measured power draw of 8 to 15 W per qubit. To understand what that figure means, it helps to look at where the industry stands today. 

The real culprit: control electronics 

Cryogenics is often singled out as the main energy drain.

Contrary to this intuition, control electronics often account for a far larger share of the total energy budget than the cryogenic infrastructure itself — including all the cables, filters, attenuators, qubit chips, circulators, and amplifiers housed inside the cryostat. 

IBM acknowledges this in its own roadmap: on its future Starling (2029) and Blue Jay (2033) systems, 80% of the power bill will come from control electronics. In other words, as quantum scales up, it is the controller — not the cryostat — that will determine whether a system remains deployable. 

That is why the quantum controller is not a secondary component. It is the component that most directly conditions the possibility of moving to industrial-scale deployment.

The numbers today

A quantum computer is not just a processor. It is a stack of interdependent layers, each drawing power. Olivier Ezratty, in his reference work Understanding Quantum Technologies (8th edition, 2025), breaks down the total power consumption of a quantum system into four main categories: 

• Cryogenics: keeping superconducting qubits near absolute zero, around 15 millikelvin 

• Control electronics: generating the microwave signals that drive each qubit and photonics/lasers for photonic, cold atoms and trapped ion qubits

• Vacuum systems: pumps and chambers 

• Classical control servers: compiling and orchestrating computations including real-time error detection

Equipping a 399-qubit system with the Chronos Q reduces total power consumption by 21% today from ~140 kW to ~110 kW.  As the industry moves toward systems with thousands of logical qubits — control electronics will represent an ever-larger share of the total energy bill, making the efficiency of the Chronos Q not just relevant, but essential.

The scaling problem 

Today's most advanced quantum systems have a few hundred to a few thousand physical qubits. Reaching genuinely useful applications — molecular simulation for drug discovery, post-quantum cryptography, large-scale combinatorial optimization — will require thousands of logical qubits, each built from millions of physical qubits. 

Ezratty's projections are concerning. Future fault-tolerant computers supporting 4,000 logical qubits could consume anywhere from a few MW to over 100 MW. PsiQuantum, for instance, estimates that its first 100-logical-qubit processor will draw a total of 100 MW — roughly half the combined power of all documented US supercomputers. For context, that is also approximately what it takes to run a mid-sized AI data center. Quantum computing, it turns out, may inherit more than just the hype. 

The conclusion is unavoidable: if control electronics do not advance radically in efficiency, quantum computing will be constrained not by qubit physics, but by available electrical power. 

Quantum vs. classical: a power bill with a different logic 

In classical computing, energy and computation are inseparable. Every operation costs electricity. 

Quantum works differently. The computation itself is essentially free. What costs energy is everything required to make it possible: keeping qubits near absolute zero, generating the precise signals that drive them, filtering every source of noise. A classical computer consumes when it computes. A quantum computer consumes to exist in a state where computing is possible. 

This is why quantum's energy problem has a ceiling that classical's doesn't. As systems scale and error correction matures, the cost of "being ready" gets amortized across an exponentially larger computational space. The qubit doesn't care how hard the problem is — it just needs to stay coherent long enough to run it. Which makes efficiency at the control layer not a secondary concern, but the central one. 

(Note: this analysis focuses on superconducting qubits, which represent the dominant architecture in today's quantum computing landscape) 

Aqsolotl's position 

The quantum boom is coming — and with it, an energy bill the industry is nowhere near ready for. Every qubit added to a system, every step toward fault tolerance, every rack of control electronics: it all adds up, fast. Aqsolotl controllers run at 8 to 15 W per qubit, regardless of system size. Within a complete room-temperature controller infrastructure, that is the most energy-efficient figure on the market today. Not a coincidence. Not a compromise. A deliberate bet on what the industry will need before it even knows it needs it.