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

[Drift Vector]

Multiple remarkably different hardware approaches, superconducting, photonic, neutral atom, all converging on roughly similar logical qubit milestones and rough timelines is either a good sign the underlying engineering challenges are becoming plainly well understood across the whole field, or it's just everyone anchoring on the same convenient, round benchmark number for marketing purposes without much independent basis behind it

[SilverSurfer51]

Worth remembering that even hitting 100 reliable logical qubits doesn't automatically translate into breaking encryption or solving really valuable industrial problems immediately. Plenty of the most consequential quantum algorithms still require considerably more logical qubits than that specific milestone number to actually run at meaningful, practical scale

[PhotonBurst48]

Curious whether Quantinuum's specific 100 logical qubit target for next year actually holds given how frequently these kinds of ambitious quantum computing roadmap announcements have slipped in the past across the whole industry generally. Would want to see the actual delivered hardware and independently verified benchmark results before treating any announced date as fully locked in

[HulkHogan_Dev]

This piece does a good job explaining why 100 error corrected logical qubits specifically has become such a widely cited milestone rather than some other arbitrary number. It's roughly the threshold where a system starts becoming useful for a truly broader range of real world applications rather than just narrow, specialized demonstrations

[PlanckLimit12]

The below threshold framing is frankly the single most important technical concept in this entire piece, more so than any specific company's roadmap or announced qubit count. Once you're reliably below threshold, adding more physical qubits especially improves things rather than just adding more places for new errors to occur, which is the actual inflection point the whole field has been chasing for years

[Sinead]

100 to 1,000 physical qubits required per single logical qubit is a quite staggering overhead ratio once you actually sit with the number properly. That completely reframes what a headline qubit count announcement from any given company actually means in terms of real, usable computational capability

[Leah]

Shifting the industry's actual benchmark from raw physical qubit count toward genuine logical qubit count is long overdue and simply should have happened even earlier in public discourse. Raw qubit announcements without corresponding error rate context have been generating misleading headlines for years now