Why Quantum Computers Aren’t Practical (Yet)

Quantum computing has captured imaginations (and venture capital) worldwide – promising to crack cryptography, simulate complex molecules, optimize logistics, and even supercharge AI. Billions are poured into qubits, and headlines abound with supposed “breakthroughs” (even solutions to Schrödinger’s cat and consciousness!). Yet in 2025, quantum machines remain largely laboratory curiosities with no clear, practical advantage over classical supercomputers. This feature article explores the long, winding road of quantum computing – its quirky history, the tangle of competing hardware technologies, and the huge gap between hype and reality. Along the way we’ll puncture some overblown claims, highlight what (if anything) quantum rigs can do today, and survey what researchers really hope to accomplish in the next 5–10 years.

A Brief (and Bumpy) History

Quantum computing may feel futuristic, but the seeds were sown decades ago. In 1980, Paul Benioff proposed a “quantum Turing machine,” laying the conceptual groundwork. Two years later, Richard Feynman famously quipped that a quantum-based computer could efficiently simulate quantum physics – a task exponentially hard for classical machines. Through the 1990s, theorists kept the fire alive. Peter Shor stunned the world in 1994 with an algorithm factoring large numbers (threatening RSA encryption) exponentially faster than known classical methods. Lov Grover followed in 1996 with a quadratic-speedup search algorithm. These results (and early error-correction schemes like Shor’s 9-qubit code) turned quantum computing from sci-fi to science.

In the late 1990s, entrepreneurs and laboratories tried building real hardware. 1999 saw the founding of D-Wave (quantum annealers), and 2000 introduced Adiabatic Quantum Computing (Farhi et al.). Over the next two decades, many prototypes emerged: 2001 – a 7-qubit NMR experiment factored 15; 2011–2014 – D-Wave installed early commercial annealers in Vancouver; 2016 – IBM launched the IBM Quantum Experience, putting a 5-qubit superconducting processor on the cloud for students and researchers to play with. 2019 brought Google’s Sycamore chip, which performed a special task in 200 seconds that Google claimed would take a supercomputer 10,000 years (a bold claim later countered by IBM researchers who suggested a few-day classical solution might suffice). Today even more qubits exist – IBM’s 1,121-qubit “Condor” chip (2023) and China’s Jiuzhang photonic system (2020) – but these still run only contrived benchmarks, not real-world apps.

Key milestones in quantum computing include:

1982: Richard Feynman proposes a quantum computer to simulate physics.
1994: Peter Shor devises his factoring algorithm (breaking RSA in theory).
1996: Grover’s unstructured-search algorithm (quadratic speedup).
1999: D-Wave Systems founded (quantum annealing device for optimization).
2001: Demonstration of Shor’s algorithm on a 7-qubit NMR QC (factoring 15).
2016: IBM releases the first cloud-accessible quantum processor (5 qubits).
2019: Google reports “quantum supremacy” on a random task (Sycamore chip).

Each step has inspired headlines – often more breathless than warranted. For example, media touted Google’s 2019 Sycamore result as “10,000 years vs 200 seconds”, but the same task was shown to be feasible on a supercomputer in a couple of days. Still, these milestones chart real progress in engineering qubits, even if practical payoffs are elusive.

A Spectrum of Qubits: Superconductors, Ions, Photons… and Others

Quantum computers come in many flavors, each with its own strengths and weaknesses. Here’s a snapshot of the major hardware platforms:

Technology	Strengths	Challenges	Examples/Companies
Superconducting	Uses well-known semiconductor fabrication; demonstrated leading qubit counts (hundreds); relatively fast gate speeds.	Requires ultracold (millikelvin) dilution refrigerators; qubits are fragile and have limited coherence (fidelity still “not up to par”).	IBM (e.g. Condor, 1121 qubits); Google (Sycamore/Willow); Rigetti; Intel
Trapped ions	Ultra-high coherence and gate accuracy (lasers control ions with “high-fidelity” operations); many ions can be nearly identical.	Complex hardware (lasers, vacuum, electromagnets); gate speeds are slower than in superconducting; scaling up to large qubit numbers is very challenging.	IonQ; Honeywell (Quantinuum); Alpine Quantum Tech.
Photonic (Light)	Operates at room temperature; photons travel fast and suffer less thermal noise (“inherent stability”); potentially easy to network.	Photon loss and limited interaction: multi-photon gates are very hard; manufacturing large, complex optical circuits is difficult.	PsiQuantum; Xanadu; Quantum Circuits Inc.
Neutral Atoms	Qubits are atoms trapped by lasers (“optical tweezers”); very promising for scaling (thousands of atoms can be trapped); low error rates.	Requires precise control of many lasers and fields; still early tech (mostly research, e.g. ColdQuanta, Harvard/Berkeley groups).	ColdQuanta; Pasqal (France).
Topological	(Theoretical) Majorana-based qubits could be inherently protected from decoherence.	Still unproven experimentally at scale; Majorana 1 chip (Microsoft) is an early attempt but “physicists not impressed” so far.	Microsoft StationQ; Rigetti (theory)

These categories barely scratch the surface (there are also spin qubits, silicon qubits, etc.), but they illustrate the diversity. No single approach has “won” – companies hedge bets across different approaches. Superconducting qubits (IBM, Google, Rigetti) have led with qubit count, but coherence times remain short. Trapped ions (IonQ, Honeywell) boast greater stability, at the expense of size and speed. Photonic startups (Xanadu, PsiQuantum) tout room-temperature operation, but struggle with photon losses and lack of universal two-qubit gates. And Microsoft’s long-anticipated topological qubit “Majorana” remains more hype than hardware, as even enthusiastic reports admit early prototypes have yet to outshine conventional qubits.

In short, each qubit type has trade-offs, and no architecture is ready to shoulder general-purpose computing yet. As one analyst quipped, “trapped-ion systems require sophisticated infrastructure such as ultra-high vacuum chambers and precise lasers,” and scaling that to thousands of qubits is daunting. Meanwhile, superconducting chips demand refrigerators colder than deep space to keep their qubits alive.

Promises vs. Reality: What Quantum Can (and Can’t) Do

The promised advantages of quantum computing are dazzling: exponential speed-ups on important problems, unbreakable encryption, ultra-fast optimization, advanced AI algorithms, and new chemical/material simulations. Popular articles describe quantum as a magic bullet for finance, logistics, drug discovery, and more. But the reality today is far more sobering.

Experts emphasize that only a few problem classes see true exponential quantum gains. As a leading quantum researcher noted, the main uses with provable exponential speed-ups are factoring large numbers (i.e. breaking RSA/ECC encryption) and simulating quantum physics (useful in chemistry and materials). Other touted applications (machine learning, large-scale optimization, finance) mostly enjoy at best polynomial (e.g. quadratic) speed-ups, which can be wiped out by quantum overhead. In practice, qubit gates are hundreds of times slower than classical operations, and quantum data input/output (bandwidth) is extremely limited. This means that for almost any problem that isn’t screamingly “quantum-native,” a classical supercomputer or GPU cluster remains faster. As one study put it, even a hypothetical quantum computer with tens of thousands of logical qubits would need centuries to outperform a top-end GPU on a real-world problem if the quantum speedup is only quadratic.

Moreover, quantum algorithms must be carefully crafted. Unlike classical programming where many problems have efficient solutions, only a narrow set of tasks (e.g. factoring, quantum simulation) truly require a quantum approach. As Scott Aaronson (UT Austin) notes, the grand claims (e.g. “quantum will revolutionize machine learning, finance, optimization… all these industries”) always demanded skepticism. Indeed, after years of NISQ-era experimentation, many lofty promises are unmet. A recent IEEE Spectrum piece bluntly observes “hype is everywhere… and practical applications are still far away”.

In practical terms, today’s quantum hardware has shown no clear commercial advantage. The Boston Consulting Group reports that to date, “quantum computing provides no tangible advantage over classical computing in either commercial or scientific applications”. Quantum devices are exceedingly noisy – qubits decohere in microseconds, and every gate adds errors – so useful algorithms (that might hide some errors) are limited to toy problems. Even Google’s Sycamore “supremacy” demonstration was on a random sampling task, not a problem with practical value. When D-Wave researchers claimed “supremacy” on a materials-science problem (spin glasses) using their annealer, critics immediately pointed out that clever classical methods could match or beat the result. In sum, quantum speed-ups remain largely theoretical. As Microsoft’s CTO predicted: “we’re decades away from demonstrating large quantum-based speed-ups” in AI or chem |🌐 (the pace of qubit errors is too high to yet tackle useful molecule designs or learning models).

Big Money, Tiny Results: The Investment Landscape

Meanwhile, the funding flood for quantum computing shows no signs of drying up. Governments and corporations worldwide are pouring money into quantum “race.” Crunchbase reports that startups raised $1.9 billion in 2024, more than double the $789 million in 2023 – despite a general tech funding cooldown. Massive rounds have become routine (e.g. PsiQuantum snagging $594M partly backed by the Australian government, Israeli Quantum Machines pulling $170M, etc.). Venture capitalists (and parents like SoftBank, Google Ventures, Intel, Microsoft) are chasing the “next big quantum success story” – and absorbing every press release of a qubit-count milestone or new chip design.

On the public side, virtually every major economy has launched quantum initiatives. Europe’s Quantum Flagship is spending €1 billion over a decade; Germany unveiled a €3 billion quantum plan in 2023; the US National Quantum Initiative directs billions to universities and labs; China claims $15+ billion in quantum tech outlay; India announced a $1B+ five-year National Quantum Mission; Israel earmarked $380M for quantum research; Singapore ~$120M; etc. In aggregate, public commitments likely top $10–15 billion globally. Corporations are also hungrily funding R&D: IBM, Google, Microsoft, Amazon, Alibaba, and many defense contractors have multibillion-dollar programs.

All this cash sends two signals: (1) investors believe quantum is a long-term “moonshot” worth funding, and (2) they expect breakthroughs enough to justify that investment. Yet the disconnect between dollars and deliverables is stark. As one consultant quipped, the quantum sector is in a “pissing contest” over claims (Google vs IBM on supremacy, or media hype around “Majorana 1”), while actual useful output is minimal. In practice, much of the quantum ecosystem today is funded for research, demonstration, and prototype hardware – essentially an exploratory R&D phase, not a full-blown industry. The Boston Consulting Group notes that despite raising record funds, “quantum computing has yet to demonstrate an advantage at scale” and error rates (fidelity) “are still not up to par”.

Governments are somewhat more patient, treating quantum as strategic R&D. But private investors – who have pumped nearly $3B in the last two years – may soon demand clearer roadmaps. Until then, the hype continues to drive headlines (and investment), even as the hardware humbly continues at the lab-bench stage.

Key Bottlenecks: What’s Holding Qubits Back?

Why exactly are qubits still such delicate flowers? The technical hurdles are formidable:

Decoherence and noise. Qubits only stay in their quantum state for a tiny fraction of a second (milliseconds or less) before environmental noise or control errors collapse them. Even a slight vibration or stray magnetic field can destroy the information. High-fidelity qubit operations require extreme isolation and precision. As a review notes, “the quantum state of qubits is very fragile and any perturbation… can affect them uncontrollably, causing the stored information to be lost”. Current devices employ dilution refrigerators at 15–20 millikelvin – colder than outer space – just to maintain qubit “ground state”.
Error rates and fidelity. Linked to decoherence, every quantum gate has only a finite chance of success. Today’s leading devices have two-qubit gate fidelities around 99%, meaning even a 50-qubit circuit can accumulate fatal errors. BCG notes “fidelity – the accuracy of quantum operations – is still not up to par and hinders broader adoption”. Without quantum error correction (QEC), even small computations fail. But QEC requires many more physical qubits per “logical” qubit (hundreds to thousands), compounding hardware demands.
Scalability and connectivity. Building a useful quantum computer might require millions of qubits (for error-corrected, fault-tolerant operation). All platforms struggle to scale. Qubit cross-talk, wiring complexity, and maintaining uniformity become nightmares as qubit counts grow. As one article quips, error-correcting thousands of qubits is akin to RAID storage: you replicate bits, but here you need “far more complicated correction codes” because qubits cannot be simply copied (no-cloning). Each new qubit added brings new sources of noise, calibration headaches, and control overhead.
Control hardware and engineering. The required equipment is massive and specialized: cryogenic dilution fridges, lasers and vacuum chambers, microwave control electronics, etc. Not only is each chip expensive and finicky, but “integrating multiple qubits, gates and components” requires bespoke engineering, precision fabrication, and automating testing at a scale that hardly exists yet. The plain-concepts analysis notes that “the required scale of cooling equipment… is beyond the feasibility of currently available equipment”.
Software and algorithms. On the software side, quantum compilers and languages are nascent. There is a “considerable lack of software” for these systems. Almost no code is portable: algorithms must be hand-tuned for each hardware’s quirks (gate set, connectivity). Enterprise software for quantum (schedulers, error mitigation, etc.) is still in research. In short, it’s easier to write a high-level program for a classical CPU than to craft a quantum circuit that actually works on today’s machines.

Each of these factors amplifies the others. Fast error-corrected quantum computing requires both far more qubits and far lower noise than we have now. Leading researchers conclude that until coherence times improve by orders of magnitude and integrated error correction is demonstrated, all but the smallest “quantum” tasks remain intractable. In the words of one scientist, “quantum computers will only really shine on small-data problems with exponential speed-ups. All the rest is beautiful theory, but will not be practical”.

Hype vs. Reality: Wild Claims and Reality Checks

If quantum researchers can be reserved, the media and PR spin cycle can be anything but. Lately it seems there’s a new “quantum breakthrough” headline every week, often bordering on science fiction. Some recent eye-catchers include:

Majorana-Magicianism: Tech magazines breathlessly hyped a so-called “Majorana 1” chip by Microsoft as a quantum breakthrough. Yet physicists were quickly quoted saying they were “not impressed” – the chip had shown nothing that rival devices couldn’t do, despite the flashy marketing.
Consciousness Solved?: A PR newswire in early 2025 trumpeted a “proof of wavefunction collapse” on a superconducting quantum computer that supposedly explained consciousness. The authors even name-dropped ChatGPT and other LLMs solving Schrödinger’s Cat. In reality, the paper was more of a sensational press release than science: “Schrödinger’s cat” is a thought experiment with no puzzle to solve, and the claim of “solving consciousness” drew obvious eye-rolls from experts.
Molecular Miracles: Startups often over-promise applications. For instance, a company might claim its 30-qubit system will “revolutionize drug discovery,” but the reality is current devices can simulate at best a handful of atoms. Notably, even Google’s team stated after their experiment, “It’s obviously a pissing contest” whether they achieved supremacy – hardly the sober language of a mature technology.
Race to Billions: In the investment press, every $100M funding round is hailed as a prophecy of imminent QC triumph. Yet the actual engineering outcomes of those billions of dollars are mostly incremental qubit-count increases and error-rate improvements. While VC money has skyrocketed (with 2024 venture funding double 2023’s), product deployments are almost non-existent. Analysts joke that right now, many quantum projects are selling the future, not the present.

These examples serve as cautionary tales: take any bold quantum claim with a grain of salt. As one IEEE Spectrum article summarizes, “the quantum computer revolution may be further off and more limited than many have been led to believe”. Skeptics note a “history of over-promise, under-deliver” is building.

What Can Quantum Do Today?

OK, so we’ve poured cold water on most grand promises. But isn’t there anything quantum machines can do right now? The answer is: yes, but mostly niche, experimental tasks. Today’s quantum computers are good at:

Quantum simulation (small scale): These machines can mimic simple quantum systems. For example, D-Wave’s quantum annealer recently ran a spin-glass simulation that classical computers struggled to handle. Similarly, IBM, Google, and academic labs routinely simulate small molecules (H₂, LiH, etc.) with variational algorithms. These proofs-of-concept are valuable research exercises but are far from industrial-strength chemistry.
Algorithm and hardware research: Universities and tech labs use current quantum hardware (often via the cloud) to test new algorithms, error-correction schemes, and hardware designs. Access programs like IBM Quantum Experience or Microsoft’s Quantum Azure let thousands of users learn by running jobs on 5–100 qubit devices. In that sense, quantum computers today are mostly R&D tools and educational instruments.
Benchmarking and PR: Many companies showcase quantum demos for publicity. For instance, a telecom firm might demonstrate an “optimized route” calculation on an annealer, or a bank might simulate a trivial option-pricing model. These exercises are more about marketing (“look what quantum can theoretically do”) than delivering a business advantage. Even if the industry benefits are minor now, some value lies in training a quantum-savvy workforce and testing cloud quantum services.
Cryptography research: Paradoxically, one “use” of current quantum efforts is to prepare for quantum threats. Organizations are actively developing post-quantum cryptography (PQC) standards, not because quantum computers have broken anything yet, but because the existence of quantum computers is driving cryptography research. Ironically, it’s now classical algorithms that solve the so-called “quantum problem” of breaking RSA.

Outside of these roles, no quantum algorithm today runs any real workload faster than classical. If your company is looking to speed up pricing models or machine learning, a GPU cluster is still the go-to solution. As Sundar Pichai (Google’s CEO) put it recently, quantum computing “is only being used for research purposes at the moment”. In practical terms, a team developing drugs or optimizing supply chains will not find their problems solved by a $15M fridge with 100 qubits – yet.

The Next 5–10 Years: Reasonable Expectations

With all that said, the field is far from static. Researchers are steadily pushing the envelope, and experts have tentative expectations for the coming decade:

Scaling up qubit counts. Major labs plan to cross new thresholds. IBM unveiled a 1121-qubit chip (“Condor”) in 2023, and promises ~4000 qubits by 2025 on its roadmap. Google continues adding qubits to Sycamore-like devices. Startups like IonQ and PsiQuantum also vow to hit the thousands-qubit mark in the next few years (often using optical technologies or multiplexing tricks). If those projections hold, 2030s machines might indeed have physical qubit counts in the tens or hundreds of thousands.
Error correction and “logical” qubits. The holy grail is fault tolerance: building a logical qubit out of many imperfect ones, so that computations can run reliably for long durations. Progress here is slower, but critical experiments are underway (e.g. surface codes, repetition codes). Experts suggest we won’t see full error-corrected quantum computers in 5–10 years, but perhaps basic demonstration of a logical qubit. Companies like Google and Microsoft pursuing topological qubits hope that even if conventional routes stall, new physics (Majorana quasiparticles) might accelerate error resistance. For now, a fully error-corrected machine remains a decade or more away for most optimistic estimates.
More algorithms (and software). As hardware improves, the hope is to discover useful intermediate algorithms. Research is active on quantum algorithms for optimization, chemistry, materials, etc. Some years from now, we might see heuristic quantum routines that slightly outperform best classical methods on specialized problems (e.g. max-SAT, portfolio optimization) – but only after significant error mitigation. In software, toolchains (compilers, transpilers, optimizers) will mature, making quantum programming more streamlined.
Narrow quantum advantage. The next practical win may come not from a universal quantum computer, but from specialized devices. For example, D-Wave’s annealers might find optimization solutions faster in narrow domains (scheduling, machine learning feature selection, materials modeling). Photonic simulators may excel at simulating boson sampling or certain quantum circuits. In other words, the first real-world “quantum advantage” applications might be bespoke and limited, rather than general-purpose.
Continued classical improvements. While all this happens, classical computing won’t stand still. Supercomputers keep breaking petaflop and exaflop barriers, GPUs and TPUs continue a rapid Moore’s-Law-like climb, and AI software efficiency improves daily. The IEEE Spectrum analysis warns that quantum is like AI circa 2010 – early, experimental – whereas classical AI (deep learning) has already gone mainstream. In short: any gains from future quantum devices will be fighting an ever-advancing classical foe.

In sum, most insiders say a “useful” universal quantum computer is still on the horizon, not immediate. Google’s Pichai predicted practical quantum in “5–10 years”; others caution 10–15 or more. For now, the community remains cautiously optimistic: expect incremental improvements (more qubits, better coherence) and maybe a few niche wins, but not a game-changing product next quarter. Meanwhile, efforts on post-quantum cryptography and hybrid quantum-classical algorithms (e.g. using quantum parts as accelerators for larger classical pipelines) will occupy researchers.

Quantum vs. Classical (and AI): A Moving Target

It’s instructive to compare quantum computing with today’s classical heavyweights – both supercomputers and AI accelerators:

Classical HPC: Supercomputers are immensely powerful and well understood. Modern GPUs (like NVIDIA’s A100/4090) deliver tens of teraflops or more and can be deployed in huge clusters (e.g. AI datacenters). They benefit from decades of engineering (error correction is trivial for RAM, and performance scales via parallelism). The BCG report notes “classical computing continues to raise the bar thanks to big strides in hardware (such as GPUs), algorithms, and AI libraries.” In fact, analyses show that even a single cutting-edge GPU can outperform early quantum machines on many tasks unless quantum has an exponential edge. Troyer’s study found that if quantum gains are merely quadratic, a future quantum chip might take centuries to beat a classical GPU on realistic problem sizes.
AI/ML workloads: The AI explosion relies on specialized silicon (GPUs, TPUs) and well-tuned software (TensorFlow, PyTorch). Quantum computers have no known advantage on typical AI tasks like neural network training or inference. Quantum data bandwidth is too slow to feed a model, and no “quantum deep learning” algorithm has materialized that outpaces classical. In fact, experts say data-intensive problems (machine learning, searching large databases) are “almost certainly out of reach for the foreseeable future”. So while quantum computing often gets lumped with AI hype, the two fields are largely separate battles: classical AI is winning now, whereas quantum is still proving it works.
Hybrid approaches: Some hopeful scenarios envision “quantum-inspired” algorithms: using ideas from quantum to improve classical computing. For instance, certain optimization or sampling techniques originally meant for quantum annealers have inspired new heuristics on classical hardware. But this is a stopgap; it means that occasionally, even a failed quantum experiment yields insights for classical methods.

In short, if you need computing power today, go classical (especially if you need deep learning or big simulations). Quantum is a very different beast – one whose promise is future advantage, not present. The competition, therefore, is to see whether quantum can keep pace with the rapid progress of classical hardware and software. So far, the classical side has held and even pulled ahead in many domains.

Unseen Hurdles: The Subtler Challenges

Beyond the obvious hardware issues, there are other less-visible reasons quantum remains immature:

Limited algorithms and problem scope. As noted, there simply aren’t that many useful quantum algorithms known. Grover’s and Shor’s algorithms are famous, but most other quantum proposals (optimization, simulation) either give at best polynomial speedups or require perfect error correction to matter. If your problem doesn’t fit a narrow mold, you likely can’t code it for a quantum machine at all.
Cryptography caveat. A common promise is “quantum will break encryption,” but that threat is very long-term. As the RSA blog explains, today’s 1000-qubit devices cannot crack 2048-bit RSA – a machine on the order of 20 million qubits would still take hours. In practice, this means data remains secure for decades, and cybersecurity emphasis has shifted to classical “post-quantum” encryption. In other words, one quantum use-case (breaking crypto) has become a driver for new classical solutions, reducing urgency for actual quantum attackers.
Resource and talent shortages. Building and programming quantum computers requires exotic expertise (quantum physics + engineering + algorithms). There simply aren’t enough quantum-savvy engineers or standardized tools. This talent scarcity slows development and adoption; companies often must build their own internal teams rather than hiring off-the-shelf talent.
Integration difficulties. Even if a fast quantum processor existed, it has to interface with the larger computing world. Integrating quantum coprocessors into data centers, developing operating systems or security for hybrid machines, and managing cryogenics in an enterprise environment are unsolved puzzles. One Gartner report warns that “integrating quantum computing capabilities into existing systems… remains a future task”.
Gig economy of hype. The field also suffers from a “publish or perish” dynamic. Because funding and attention are tied to breakthroughs, teams may oversell incremental results. Jurjen Bos’s recent analysis showed how almost every week a quantum group claims a “breakthrough,” often with little substance. This hype culture can lead to public disappointment when claims fall short (e.g. unfulfilled promises of “magical” qubit materials).

All these factors combine into a complex reality: quantum computing isn’t just one unsolved problem, but a stack of them – physics, engineering, software, infrastructure, and even public perception. Until a significant fraction of these challenges are overcome, the “Practical Quantum Computer” remains more of a slogan than a deliverable product.

Wrapping Up: Patience and Perspective

In closing, quantum computing is an extraordinary technology in principle, and it will likely revolutionize some fields eventually. But as of 2025, it is still mostly a work in progress. The gap between hype and real-world impact is wide:

We have billion-dollar investments and grand proclamations, yet only rudimentary devices at hand.
Historical milestones (Feynman, Shor, Grover, Google supremacy) show potential, but each milestone also highlights new obstacles.
Hardware types (superconducting, ion, photon, etc.) all make progress, but none have swept all others away or conquered noise and scale.
Many exciting promised applications remain out of reach without major leaps in error correction and qubit quality.

Think of today’s quantum computers like experimental race cars on a test track. They rev high with theoretical potential (exponential speed-ups), but they keep stalling, spinning wheels (errors), and needing pit stops (cooling). Meanwhile, the highway of progress on the broader computing “autobahn” remains crowded and fast-moving with classical HP and AI “vehicles.”

Yet it’s not all doom and gloom. Quantum research is yielding new physics insights, training a generation of quantum engineers, and spinning off smaller gains (quantum-inspired algorithms, better simulators). In 5–10 years we may see the first applications where a quantum device outperforms any other option (likely niche, highly specialized tasks). Companies will learn which problem areas truly benefit from quantum style computing.

For now, however, quantum computers are useful mostly as research and educational tools – and sometimes as PR assets – rather than as workhorses of industry. As two prominent voices put it: “Quantum computing is still in its infancy”, and achieving its legendary promises will take decades and concerted effort. The field needs patience, realism, and sustained support to cross from fascinating experiments to practical technology.