Quantum scaling should focus on compute-per-watt, not qubit count

TechRadar's Matt Rijlaarsdam argues that quantum computing should be measured by useful computation per watt rather than qubit count alone.

The article notes that UN researchers expect artificial intelligence to double the power and water used by data centers by 2030. It says power availability and total cost of ownership will remain central as computing demand grows.

Superconducting quantum processors have remained at around 100 qubits for almost a decade, the article says, because more than 90% of current chips' surface is used for wiring that controls and reads the qubits. Connecting many smaller processors can raise qubit counts, but networking losses and sparse connections can reduce compute-per-watt.

The article estimates that, at today's price per qubit, a million-qubit machine would cost between 100 billion and 1 trillion dollars. The cost per qubit would need to fall at least 100-fold for quantum computers to become economically viable, it argues.

Rijlaarsdam points to volume manufacturing, standardized parts and an open architecture—called Quantum Open Architecture—as potential ways to reduce costs. The article says processors, cryogenics and control electronics could come from specialist companies. It concludes that building the supply chain may take years, if not decades, and that choices made in the next few years will shape whether quantum computing delivers useful systems.