Quantum Computing in 2026: Which Problems Are Actually Solvable Today vs. Still Theoretical?

Most coverage of quantum computing in 2026 falls into one of two failure modes: uncritical hype or dismissive scepticism. The truth requires distinguishing between what hardware can do today and what algorithms actually require.

## The Hardware State in 2026

IBM's quantum roadmap has been the most publicly documented progression. The company moved from the **Eagle processor** (127 qubits, 2021) to **Osprey** (433 qubits, 2022), **Condor** (1,121 qubits, 2023), and the **Heron** generation — focused on reduced error rates rather than raw qubit count. The key insight: raw qubit count is a misleading metric.

What matters is the **error rate per gate operation** and the **coherence time** — how long a qubit maintains its quantum state before decoherence destroys the computation. Current best-in-class superconducting qubits achieve gate fidelities of 99.5–99.9% for two-qubit gates. This sounds high. For a 1,000-qubit computation requiring 10,000 gate operations, the probability of at least one error is nearly 100%. Fault-tolerant quantum computing requires error rates below **10⁻⁴ per gate**, achieved through quantum error correction codes that encode one logical qubit into hundreds of physical qubits. We are not there yet at scale.

Google's **Willow chip** (2024) demonstrated a genuine milestone: error rates that decrease as more qubits are added — below the surface code threshold. This was a real technical achievement. It is not a demonstration of useful computational advantage over classical computers for practical problems.

---

## What Quantum Computing Can Actually Do Today

**Quantum simulation** is the application with the strongest near-term case. When you want to simulate quantum mechanical systems — molecular energy levels for drug discovery, material properties for battery chemistry, catalyst behaviour for industrial chemistry — you are using a quantum computer to simulate nature's own computation. IBM's collaboration with Daimler on lithium-air battery molecular simulation demonstrated chemistry insights inaccessible to classical computation at the scale modelled.

**Quantum annealing** (D-Wave's approach) has been commercially deployed for constrained optimisation problems: portfolio optimisation, traffic flow, logistics routing. D-Wave's Advantage system provides genuine value in specific formulations where classical heuristics are computationally expensive. The limitation: only a subset of optimisation problems map naturally to the Ising model topology.

> ⚡ Quantinuum's H2 trapped-ion processor achieved a record circuit depth of 100 with 56 qubits and near-full connectivity — demonstrating that trapped-ion systems, despite lower qubit counts than superconducting chips, may reach commercial utility first due to higher per-qubit fidelity.

---

## What Quantum Computing Cannot Do Today

**RSA encryption breaking** requires running Shor's algorithm at scale — estimated to require millions of error-corrected logical qubits to break RSA-2048 within hours. This is 15–20 years away at minimum under the most optimistic hardware scaling assumptions. The "store now, decrypt later" threat to current communications is real but requires an adversary with the patience to store encrypted traffic for two decades.

**Broad commercial advantage** over classical computing does not yet exist. For most machine learning, data analytics, and optimisation tasks, well-optimised classical algorithms on modern GPUs outperform current quantum hardware in both speed and cost. This is not a permanent state — it is a statement about where the hardware is today.

| Capability | Today | 5–10 Years | 15+ Years |
|-----------|-------|-----------|---------|
| Quantum simulation (chemistry) | Limited commercial use | Broad drug discovery | Mature |
| Optimisation (annealing) | Niche commercial use | Expanded applications | — |
| Machine learning speedup | Not demonstrated | Uncertain | Possible |
| RSA-2048 breaking | Not possible | Not possible | Possible |
| Broad classical computing replacement | Not possible | Not possible | Speculative |

---

## The Metrics That Matter: Quantum Volume vs. Qubit Count

IBM's **Quantum Volume** metric attempts to capture the useful computational power of a quantum system more honestly than raw qubit count. It accounts for qubit count, connectivity, gate fidelity, and coherence time in a single number. More recently, IBM introduced **algorithmic qubit** measures — how many qubits can run a useful circuit without exceeding 1% error probability.

Quantinuum's H2 achieves a Quantum Volume of 4,194,304 (2²²) with 56 qubits. Google's Sycamore has high qubit count but lower Quantum Volume per qubit due to limited connectivity. The trapped-ion versus superconducting comparison shows that architecture choice involves fundamental tradeoffs between qubit count, connectivity, and gate fidelity that cannot be optimised simultaneously with current technology.

## The Bigger Picture

Quantum computing in 2026 is a technology in genuine transition — past pure laboratory curiosity, not yet at reliable commercial utility for most applications. The timeline for broad commercial advantage is a decade minimum, not years. The companies best positioned are those building quantum-classical hybrid algorithms that leverage quantum hardware for the specific subroutines where it provides advantage — not those betting on quantum computing replacing classical computing wholesale.

The honest investment thesis: quantum simulation for chemistry and materials is the near-term commercial application most likely to produce defensible ROI. Cryptography disruption is a long-horizon risk requiring dedicated monitoring but not immediate alarm. Everything else is development infrastructure for capabilities that do not yet exist at scale.

Quantum Computing in 2026: Which Problems Are Actually Solvable Today vs. Still Theoretical?

// COMMENTS

ON THIS PAGE