A leading quantum analyst says the hype is real, but so is a coming shakeout of 85 hardware companies

[StoneCold]

Bob Sorensen, chief analyst for quantum computing at Hyperion Research, laid out a genuinely useful reality check on the current state of the industry in a recent Catalyst podcast interview, separating real progress from the artificial benchmarks that keep getting mistaken for it

His clearest example of misleading hype is boson sampling, a demonstration where quantum systems simulate something like a pachinko machine, tracking how photons bounce and scatter in ways that are genuinely intractable to simulate classically. One company claimed a speedup of 10 to the 28th power over classical computers and argued this proved quantum computing operates across multiple universes. Sorensen's blunt response, engineers designing aircraft or running crash tests don't want to hear that multiple universes are required to do useful computing, since the demonstration has no practical application whatsoever, it was built purely to be easy for quantum and hard for classical, not to solve a problem anyone actually needed solved

The real bottleneck right now is error correction, not raw qubit counts. Current hardware sits in what's called the noisy intermediate scale quantum era, meaning you don't run a quantum algorithm once and get an answer, you run it a thousand times and hope the correct answer appears in 70 to 80 percent of the resulting histogram. A decade ago it took thousands of error prone physical qubits to build just one reliable logical qubit, and getting that ratio down is what actually determines whether a system is useful, not the headline qubit count companies advertise. Sorensen puts genuinely useful fault tolerant systems three to four years out, once the industry can reliably turn something like a million physical qubits into thousands of dependable logical ones

He's notably candid about where AI and quantum actually differ for scientific discovery. AI is data driven, essentially making its best guess based on existing known information, and its conclusions are opaque, you can't fully explain why a model reached a particular answer or reliably reproduce it. Quantum simulation is grounded directly in physics rather than pattern matching against a training set, meaning its results are explainable and reproducible in a way AI's aren't, making the two genuinely complementary tools rather than competitors, AI helps guide which experiments are worth running, quantum actually simulates the underlying physical reality

Sorensen's biggest concern isn't the technology's trajectory, it's the roughly 85 companies currently trying to become quantum hardware vendors, a number he says has no precedent in the history of computing, there were never 85 companies making mainframes, PCs or smartphones. He expects 70 of them could fail within two years without meaningfully damaging the sector's real prospects, but worries that wave of failures gets misread by investors and governments as proof the whole field is another AI winter in the making, rather than the normal, healthy consolidation any young industry eventually goes through

When will quantum computing have its breakout moment?www.latitudemedia.com

[HeartbreakKidStinger64]

The pachinko machine explanation for boson sampling is such a clear, almost embarrassingly simple way to describe what a lot of these headline benchmark claims actually amount to underneath the hype

[Layla81]

85 hardware companies with no precedent anywhere else in computing history is a genuinely striking number, feels obvious in hindsight that a real consolidation wave is coming regardless of how good the underlying technology actually is