The race to 100 error corrected qubits is becoming the real measure of quantum computing progress

[Pat82]

A Bloomberg Opinion column this week argues that the quantum computing industry's actual progress is increasingly being measured not by raw qubit counts but by how many error corrected logical qubits a system can reliably sustain, with 100 logical qubits emerging as a widely cited threshold companies are actively racing toward. Quantinuum has specifically stated it plans to release a computer next year with 100 error corrected qubits, joining a broader industry pattern where multiple companies pursuing genuinely different underlying hardware approaches are converging on roughly similar near term milestones.

The column frames this moment against the sweep of computing history more broadly, recalling how the Eniac, built in 1946 as the first programmable general purpose digital computer, weighed about 30 tons, filled an entire room with 18,000 vacuum tubes, cost roughly 9 million dollars in today's money, and could perform only about 5,000 additions per second. A typical modern laptop, by contrast, can now perform more than 3 billion calculations per second and costs around 1,000 dollars, illustrating just how dramatically classical computing scaled once its own foundational engineering challenges were actually solved.

That historical comparison sets up the column's underlying argument about why the shift toward tracking logical qubits specifically, rather than raw physical qubit counts, matters so much right now. A quantum computer with a thousand physical qubits might only actually deliver somewhere between 10 and 100 particularly error corrected logical qubits, and possibly far fewer if error rates run high or the underlying error correction code proves inefficient, since breaking a physical qubit's inherent noise down into a reliable logical unit typically requires anywhere from 100 to 1,000 physical qubits for every single logical qubit produced. That's exactly why recent demonstrations from companies like Google and IBM increasingly emphasize whether logical error rates actually decrease as more physical qubits get added, a property known as operating below threshold, rather than simply publishing an ever larger raw qubit count on its own.

Different companies are pursuing meaningfully different roadmaps to reach comparable milestones. IBM's Quantum Starling system targets 200 logical qubits by 2029 capable of executing 100 million error corrected operations, eventually aiming for 1,000 logical qubits in the early 2030s using quantum low density parity check codes designed to reduce the physical qubit overhead per logical qubit substantially. Photonic and neutral atom approaches are pursuing broadly similar logical qubit targets on their own separate timelines, meaning the race toward 100 reliable error corrected qubits has effectively become the shared, cross platform benchmark the whole industry is being measured against, regardless of which specific underlying hardware technology any individual company has actually bet on

Quantum Computing The Race To 100 Error Corrected Qubits Is Defining Progresswww.bloomberg.com

[Grim Callum]

The Eniac comparison is a distinctly effective way to frame just how early quantum computing still actually is relative to where classical computing eventually ended up. Nobody in 1946 could have predicted a 1,000 dollar laptop doing 3 billion calculations per second, and it's a useful humility check for anyone trying to confidently predict where quantum computing sits in another 80 years