GoKawiilI’ve been told that I’m a good guesser. For quantum technologies, you have to be, because the bets are long term. The best guesser I ever met was my first boss, Jag Shah, who ran quantum-research programmes at the US Defense Advanced Research Projects Agency (DARPA) in Arlington, Virginia. He hired me in 2007 and asked me to write down every miracle needed to build a quantum computer.
How academics and ‘big tech’ can learn from one another
Build a quantum computer, and it could design molecules and materials, model the reactions that feed and power the world and answer questions that no classical machine can — such as how electrons behave inside high-temperature superconductors. That is the prize.
Twenty years on, here is my guess. In five years or less, a quantum computer will do something that someone needs and no ordinary computer can do, with a solution that changes how we look at the world. Give it a few more years, and a wholly new form of computing could be deployed worldwide. Quantum technology can spread that fast because it builds off investments in chips and artificial intelligence. You reach it through the ‘cloud’, and when it works, it works everywhere.
For that to happen, everyone must play their part. Governments need to fund long-horizon science that no company will pay for and help to identify what is worth buying. They must become demanding customers, in fields ranging from scientific discovery to defence. Companies speed the process along. And universities train the next generation.
Why even physicists still don’t understand quantum theory 100 years on
None of this is easy. Quantum computers run on qubits (bits of quantum information), and the hard part is not the number needed, but how to control them. Quantum computers need to manage tens of thousands to millions of qubits at once, each with its own wiring. Your laptop’s data encoding is error-free; a quantum computer’s is not. The cure is quantum error correction: redundancy that builds a few reliable qubits from many shaky ones.
The silicon processor described in this issue of Nature shows how far we’ve come (Members of the HRL Quantum Team and collaborators. Nature 655, 1154–1159; 2026). Its encoded qubits are actively error-suppressed, run by a single cryogenic control chip and connected by a superconducting ribbon cable. It grew from seeds planted years ago at DARPA. In those days, we wrapped fittings in Teflon tape and tied down wiring with dental floss — a ‘duct tape’ solution for isolating qubits in temperatures colder than those in deep space.
I’m still surprised by the bets that paid off, and by the ones I was sure wouldn’t. Neutral atoms and photons, for example, were left for dead because of scalability concerns — until academia revived them and the market embraced them. The leading technology candidate keeps changing.