The Quantum Computer Dream Is Getting a Reality Check

01 THE HANGOVER AFTER THE HYPE

A decade of quantum computing promises is colliding with an uncomfortable decade of engineering reality. Theoretical physicist Sabine Hossenfelder, one of the field most persistent skeptics, recently published a widely viewed assessment arguing that the gap between quantum marketing and quantum hardware has stopped closing on schedule. Her argument deserves engagement rather than dismissal, because the underlying numbers are checkable.

The core tension is simple. Quantum computers promise dramatic speedups for specific problems, including factoring and simulating molecules, but only if the machines can run computations long enough without decohering. The industry roadmap to that point has repeatedly slipped, and 2026 finds the field in an uncomfortable middle: too much progress to dismiss, too little to declare victory.

Chart 1: Announced physical qubit counts for landmark superconducting processors. IBM deliberately shrank from Condor to Heron to prioritize qubit quality over count. Source: company announcements via Wikipedia.

02 QUBITS ARE NOT THE METRIC THAT MATTERS

The number most often quoted in press releases, the physical qubit count, is nearly meaningless on its own. A qubit is only useful if it can hold its state, be entangled with its neighbors, and be measured on demand, and the error rates for those operations are what actually gate progress. IBM demonstrated the point in reverse when it followed its 1121-qubit Condor with Heron, a chip with roughly one-eighth the qubits but dramatically better quality, because the industry had quietly conceded that raw count was a marketing number.

A useful quantum computer needs logical qubits, error-corrected combinations of many physical qubits that behave like one reliable unit. Until recently, building a logical qubit required somewhere near a thousand physical qubits, a ratio that made useful machines a distant fantasy. That ratio has begun to fall, which is the real story of the last two years.

Chart 2: The falling ratio of physical qubits per reliable logical qubit. Willow demonstrated error rates improving as the code grew, the first time scaling helped instead of hurting. Source: Google Quantum AI, Nature (2024).

03 WILLOW AND THE FIRST REAL CRACK

Google Willow chip, a 105-qubit superconducting processor announced in 2024 and published in Nature, delivered the field first genuine crack in the error-correction wall: as the error-correcting code grew larger, the error rate went down instead of up. Passing that below-threshold milestone means that scaling now helps rather than hurts, and it converted the physical-to-logical qubit ratio from a fixed tax into an engineering curve that can be pushed. Willow cannot run anything commercially useful, but it moved the central obstacle from theoretical to merely very hard.

04 WHAT THE SKEPTICS GET RIGHT

The skeptical case rests on three observations that hold up. First, no quantum computer has yet delivered a commercially valuable computation that a classical machine cannot match, and classical algorithms keep improving in response. Second, timelines sold to investors routinely omit the error-correction overhead, presenting physical qubit counts as if they were logical ones. Third, the best-known speedups apply to narrow problem classes, while the general-purpose quantum laptop of popular imagination remains physically implausible. Hossenfelder point is that these three facts together justify considerably more humility than the average funding pitch contains.

05 WHERE THE FIELD IS ACTUALLY PROGRESSING

Genuine progress is concentrated in three areas: error-correction research, which Willow anchors; quantum simulation of small molecules and materials, which is the application with the clearest physical justification; and hybrid algorithms that combine classical and quantum components so each does what it does best. IBM and Google both publish roadmaps pointing to the early 2030s for machines with enough logical qubits to attempt useful chemistry, which is a more honest forecast than the startup timelines that promise utility next year.

06 THE 2026 SYNTHESIS

The defensible position between the boosters and the skeptics is this: quantum computing is real physics with one verified milestone of scaling error correction, a plausible path to narrow usefulness within a decade, and no credible path to broad disruption anytime soon. Investors and policymakers should treat breathless claims about quantum supremacy the way they treat pre-revenue biotech press releases, as enthusiasm requiring verification. The dream is not falling apart, but it is being forced, qubit by corrected qubit, to earn every step.