IBM says quantum advantage is three years away. They've been saying that for ten years. Google claims quantum supremacy milestones that don't translate to practical problems. Startups raise billions based on algorithms that may never run on hardware that may never exist. The quantum computing industry has perfected the art of promising revolution while delivering research. It's time to separate the physics from the marketing.

The Hype Cycle

Quantum computing has occupied a unique position in technology hype. It's simultaneously imminent and distant, revolutionary and incremental, inevitable and uncertain. This ambiguity has persisted for decades, sustained by genuine scientific progress that never quite reaches commercial relevance.

The pattern is familiar. Every few years, a major milestone generates headlines: a new qubit record, a new algorithm, a new claim of quantum advantage. The press coverage follows a script — revolutionary potential, timeline estimates, cautious caveats. Then the cycle repeats without commercial products materializing.

IBM's public roadmap has been particularly illustrative. In 2016, they predicted quantum advantage by 2020. In 2020, they predicted it by 2023. In 2023, they predicted it by 2026. Each prediction accompanied by genuine technical progress — more qubits, better error rates, new architectures — but never by the breakthrough that would make quantum computers practically useful for commercially relevant problems.

This isn't deception. It's the nature of pushing fundamental research toward application. The milestones are real but partial. The progress is genuine but insufficient. The timelines are optimistic but not dishonest. The result is a continuous stream of almost-breakthroughs that maintains investment without delivering products.

What Quantum Computers Actually Are

To understand why quantum computing remains pre-commercial, you need to understand what quantum computers actually do. They exploit quantum mechanical properties — superposition and entanglement — to process information in ways classical computers can't efficiently replicate. This isn't just faster computing; it's different computing, useful only for specific problem classes.

The key phrase is "specific problem classes." Quantum computers excel at problems with particular mathematical structures: factoring large numbers, simulating quantum systems, solving certain optimization problems. They offer no advantage for general computing tasks — word processing, web serving, database queries, AI training. The problems where they excel are important but limited.

Factoring is the famous example. Shor's algorithm, developed in 1994, showed that quantum computers could factor large numbers efficiently. This threatens RSA encryption, the foundation of modern secure communications. The threat is genuine — a sufficiently powerful quantum computer could break current encryption. But the "sufficiently powerful" requirement remains distant.

The numbers are stark. Breaking current RSA keys requires roughly 20 million physical qubits with current error correction schemes. Today's largest quantum computers have around 1,000 qubits. The gap isn't closeable through incremental improvement — it requires orders-of-magnitude advances in hardware, error correction, and algorithm efficiency.

This gap is often obscured by announcements of smaller factoring demonstrations. IBM or Google will factor a 48-bit number, generating headlines about quantum threats to encryption. But 48-bit numbers are trivial for classical computers. The demonstration proves the algorithm works, not that it's practical. The gap between proof-of-concept and practical capability remains enormous.

The Error Problem

Quantum computers are fundamentally analog devices operating in a noisy environment. Qubits are delicate quantum states that decohere — lose their quantum properties — through interaction with their environment. This isn't a design flaw; it's physics. Maintaining quantum coherence requires extreme isolation from noise.

The error rates are currently around 0.1% to 1% per operation. This sounds manageable until you consider that quantum algorithms require millions or billions of operations. Error rates compound, making long computations unreliable without error correction.

Error correction is the field's most active research area, and also its most sobering. To perform reliable computation on noisy hardware, you need logical qubits — error-corrected qubits — built from many physical qubits. Current estimates suggest you need 1,000 to 10,000 physical qubits per logical qubit, depending on the error correction scheme and the required computation length.

For the 20 million qubits needed to break RSA, you might need 20 billion physical qubits with current error correction. Today's hardware has around 1,000 physical qubits total. The gap isn't years away; it's decades away, assuming current improvement rates continue.

Error correction schemes are improving. New approaches promise better efficiency. But even optimistic projections suggest that practical, error-corrected quantum computers for cryptographically relevant problems remain 15-20 years away. Shorter timelines require breakthroughs that haven't materialized and may not exist.

Where Quantum Computing Actually Helps

The honest assessment: quantum computers are useful now for a very narrow set of problems, and this set isn't expanding as fast as hoped. The useful applications fall into three categories.

First, quantum simulation. Simulating quantum systems — molecular behavior, material properties — is naturally suited to quantum computers because they are quantum systems. This is the most mature application area, with demonstrated advantages for specific chemistry and materials problems.

But the advantages are narrow. Current quantum computers can simulate small molecules and specific quantum phenomena. The commercially relevant simulations — drug discovery, catalyst design, novel materials — require larger systems than current hardware can handle. The trajectory is positive; the timeline is long.

Second, optimization. Certain optimization problems have structure that quantum algorithms can exploit. The most prominent is quantum annealing, used by D-Wave systems, which addresses specific combinatorial optimization problems. The advantages here are contested — whether quantum annealing provides genuine quantum speedups or just efficient classical approximation remains debated.

Third, machine learning. Quantum machine learning is an active research area with theoretical potential. But current quantum computers lack the scale and coherence to demonstrate advantages on real datasets. The field remains promising but unproven.

What's notably missing from this list: cryptography, general AI, database search, web services, most of what we use computers for. Quantum computers will not make your laptop faster, your cloud cheaper, or your AI training quicker. They're specialized devices for specialized problems, and the specialized problems where they work are fewer than advertised.

The Commercial Reality

Quantum computing companies have raised billions based on future potential. The business model is largely research funding rather than product revenue. This isn't inherently illegitimate — fundamental research requires investment — but it creates incentive structures that favor continued promises over honest assessment.

The few commercial applications generating revenue are narrow. D-Wave sells quantum annealers for specific optimization problems — logistics, scheduling, portfolio optimization. The advantages are modest and contested. Customers are often research collaborators as much as commercial users.

IBM, Google, Microsoft, and Amazon offer cloud access to quantum computers. The customers are researchers, startups, and curious enterprises exploring the technology. Revenue is negligible compared to classical cloud services. The business model is preparation and learning, not practical deployment.

The honest question for commercial buyers: what problem do you have that quantum computers can solve better than classical alternatives? For most organizations, the answer is currently "none." The problems where quantum advantages exist are specialized research problems, not business operations.

This will change. The technology is improving. But the timelines for commercially relevant applications are measured in decades, not years. Organizations preparing for quantum threats to encryption should monitor developments, but the immediate risk remains distant enough that cryptographic migration timelines can be measured in years rather than months.

The Investment Question

Quantum computing is legitimately important fundamental research. It's not clear that it's a good investment target for commercial returns on currently imagined timelines. The physics problems are hard. The engineering problems are harder. The economic problems — finding commercially relevant applications at current capability levels — are perhaps hardest.

For enterprises, the appropriate stance is informed monitoring, not urgent preparation. Track the technology's development. Understand the problem classes where advantages may emerge. Maintain awareness of cryptographic implications. But don't redirect significant resources based on near-term quantum advantages that are unlikely to materialize.

The quantum computing community will continue announcing milestones. Some will be significant. Many will be incremental progress packaged as breakthrough. The challenge for observers is distinguishing genuine progress from continued promises — maintaining hope without abandoning critical assessment.

Quantum computers are real. They're getting better. They're not revolutionizing computing on any timeline that matters for current business decisions. The reality check isn't pessimism; it's honesty about where the technology actually stands and what it can actually do today.