Atom Computing CEO says quantum utility needs speed, not just more qubits

[MessiGOAT48]

Quantum World Congress has published a summary of a session with Ben Bloom, founder and chief executive of Atom Computing, and his message fits nicely with the Q-Factor piece from yesterday. Bloom's argument is that useful quantum computing needs more than large qubit counts. It needs speed, along with connectivity, fault tolerance, custom control hardware, software and networking all working together. A huge machine that runs too slowly will not solve real problems in a useful time

Speed is a known weak spot for neutral atom systems, which is what Atom Computing builds. Neutral atoms are great for scale, since you can trap thousands of identical atoms with lasers, but operations and readout tend to be slower than superconducting chips. Error correction makes that worse, because you need to run many rounds of checks quickly. If each round is slow, the whole computation drags. Every round of checks needs atoms imaged and reset

Atom Computing has been one of the leading neutral atom companies for a while. It passed the 1,000 qubit mark back in 2023, has demonstrated error correction with neutral atoms, and is working with Microsoft on a commercial machine with logical qubits. In June the company said its total funding had passed $300 million, including a $100 million Series C led by Third Point Ventures and a planned $100 million commitment from the US Department of Commerce. It also works with DARPA's Quantum Benchmarking Initiative and Nvidia

It is interesting to see a neutral atom company being so open about the speed problem rather than just talking up qubit numbers. Custom control hardware is probably where most of the gains will come from. Faster lasers, quicker imaging and better electronics could all shrink the time per operation. Whether they can close the gap with superconducting systems is the big question

Between this and the Q-Factor article, there seems to be a shift in how quantum firms talk. The conversation is moving from how many qubits to how useful those qubits are. Is speed the right thing to focus on? Or are fidelity and error rates still more important?

Atom Computing Says Utility Requires More Than Qubits. It Requires Speed. | Quantum World Congresswww.quantumworldcongress.com

[Kane_Node]

Fidelity still matters more to me. A fast machine with high error rates is useless, because errors pile up faster than you can fix them. Speed only helps once the qubits are good enough

[Paul_26]

Custom control electronics is an underrated area. Everyone focuses on the qubits, but the classical hardware around them decides much of the performance. Fast lasers and cameras are hard to source off the shelf

Building them in house gives a real edge. It also costs a fortune. Only the best funded firms can manage it

[Zidane]

The Microsoft partnership gives Atom Computing a lot of credibility. Microsoft does not put its name on systems lightly, especially after its own topological qubit troubles. Having a big cloud partner also solves the distribution problem. It is a good fit for both sides

[DarkMatter92]

Nice to see a CEO talking openly about a weakness. Most quantum companies only discuss the numbers that make them look good. Bloom deserves credit for this

[ThreadNecro98]

Speed matters because of error correction. You might need millions of rounds of checks for one useful calculation

If each round takes a millisecond, you are waiting a long time. Speed is not a luxury. It decides what is practical

[TheRizz]

Department of Commerce money is a sign of how strategic quantum has become. Governments want domestic champions

Atom Computing seems to be one of the American firms they are backing. The CHIPS Act showed how far Washington will go for strategic industries