For Quantum, Change the Ruler

A friend who runs long-term capital asked us: quantum computing — can we look at it now?

The view up front. You can, but you need a different ruler. The old ruler measures the spectacle: who has more qubits, whose numbers look brightest, whose papers make the most noise. The new ruler measures one thing only: the engineering of error correction. Why? Because until error correction is genuinely solved, the longer it computes the more it gets wrong. The commercial value does not stand up.

Where the First Sentence Lands Tells You Whether They Understand

We once heard a cutting remark: anyone who opens by talking about algorithms and software probably does not understand quantum. Cutting, but fair. The bottleneck in quantum is manufacturing, not the model. Which error-correcting code you choose decides what the chip looks like, how the wiring runs, how the qubits connect, how the control and measurement electronics are matched. Below that it pulls in process: from picking chips one at a time in the lab towards wafer-scale volume. Software ships a version every six months; a production line takes five years a generation — a tenfold difference in patience. So when we listen to someone talk about quantum, we listen first for where it lands.

A rough calculation first (figures illustrative). Error correction trades quantity for quality: dozens of physical qubits for one usable logical qubit. Every ten per cent cut from that ratio, the line's yield has to rise a notch. In other words, the real denominator is not the total number of qubits, it is how many process steps, control channels and wiring runs one logical qubit has to absorb. The denominator is written on the production line.

So which projects are the most dangerous? Not the ones that are too early, but the ones using the wrong metric to prove they are not early. Anyone who only cites single-point metrics and cannot explain the error-correction path and the manufacturing path gets marked down. This industry is moving from the scientist's bench to the engineer's systems campaign. First, is the route right; second, is the hardware-and-software system complete. Apply both and the list gets much shorter. We are fairly sure of this one.

Understanding It Is Not the Same as Being Able to Invest

The other half has to be said too. A sound technical judgement and a payoff window that lines up are two different things. On frontier lines like quantum computing and controlled fusion, the pendulum of commercialisation often swings longer than a fund life: a term of about ten years against a judgement that only resolves in fifteen. So some firms say they understand it and still will not invest. Is that a lack of belief? No, it is honesty. A firm's limit is written in its fund life, not in its understanding.

Our own arithmetic is a little different. So we can sit on the bench. But sitting still is not the same as doing nothing.

A Ten-Year Line Starts Being Drawn Today

Quantum computing is the third line in our direction matrix: commercial space, quantum computing, controlled fusion, brain-computer interfaces. The other four are AI foundation models and wholly new paradigms; physical AI, embodied intelligence and robotics; AI infrastructure; and intelligent manufacturing and hard-tech products that can compete globally. Five directions, one chain.

So what have we been doing these past months? Turning ourselves into half-insiders. Reading technical roadmaps, visiting the people who do process and control electronics, then explaining the industrial logic point by point to long-term capital that is willing to wait. Understand, match, close — the order cannot be reversed. The above is our view at this moment and does not constitute investment advice. A ten-year line has to start being drawn today.