Investor's Guide
Understanding Quantum Computing
A plain-language guide to the science, the hardware race, and what it means for investors and professionals.
The Basics — What Is Quantum Computing?
Quantum computing has attracted over $40 billion in investment since 2020. But most coverage oscillates between breathless hype and dismissive skepticism. This guide explains what's actually happening, why it matters, and how to think about the companies building it.
Bits vs Qubits
Classical computers use bits — each is either a 0 or a 1. Qubits can exist in superposition, representing both 0 and 1 simultaneously during computation. Think of a classical bit as a coin lying flat: either heads or tails. A qubit is a spinning coin — it's neither heads nor tails until it lands (is measured). This lets quantum computers explore many possibilities at once.
Superposition
A qubit in superposition isn't literally "both 0 and 1 simultaneously" — that's the common misconception. It's in a probabilistic state that collapses to a definite value on measurement. What makes it powerful: n qubits can represent 2n states simultaneously during computation. Ten qubits explore 1,024 states at once. 300 qubits represent more states than there are atoms in the observable universe.
Entanglement
Two qubits can be entangled— correlated such that measuring one instantly determines the other's state, regardless of distance. This isn't faster-than-light communication; no information is transmitted. It's a computational resource. Einstein called it "spooky action at a distance." It's real, experimentally confirmed, and central to many quantum algorithms.
Interference
Quantum algorithms are designed to amplify computational paths that lead to correct answers and cancel paths that lead to wrong ones — like noise-canceling headphones for probability. This is the actual mechanism by which quantum computers solve problems, not brute-force parallelism.
Measurement and Decoherence
Qubits are fragile. Interaction with the environment — heat, vibration, electromagnetic noise — causes decoherence: the quantum state collapses prematurely, destroying the computation. This is the central engineering challenge of the field. Most quantum computers today operate at temperatures colder than outer space to minimize it.
What Can Quantum Computers Actually Do?
Correcting hype without dismissing genuine progress.
Where quantum excels
What quantum will not do
The Hardware Race — Six Approaches
The quantum industry is not converging on a single technology. Six distinct physical implementations are in active competition.
Superconducting Qubits
Most MatureThe dominant approach today. Used by IBM, Google, and most major cloud providers.
Electrical circuits cooled to ~15 millikelvin (colder than deep space) exhibit quantum behavior. Microwave pulses manipulate qubit states. The technology leverages decades of chip manufacturing expertise.
Pros
- +Fast gate speeds (nanoseconds)
- +Compatible with existing chip manufacturing
- +Most mature ecosystem and tooling
- +Largest demonstrated qubit counts
Cons
- −Requires extreme cooling — bulky, expensive cryostats
- −Short coherence times (microseconds)
- −Scaling wiring is a major bottleneck
Investor note: Most funded approach globally. IBM and Google have the deepest moats. Watch IQM for European HPC integration — they're embedded in national supercomputing centers.
Trapped Ion
Most MatureCompeting with superconducting for leadership. Best gate fidelity of any modality today.
Individual atomic ions are suspended in electromagnetic fields and manipulated with precise lasers. The ion itself is the qubit — no fabrication required. Qubits are identical by nature.
Pros
- +Highest gate fidelity (99.9%+)
- +Long coherence times (seconds to minutes)
- +All-to-all qubit connectivity
- +No qubit-to-qubit variation
Cons
- −Slower gate speeds (milliseconds vs nanoseconds)
- −Laser systems are complex and expensive
- −Scaling to thousands of qubits is unsolved
Key Companies
Investor note: Quantinuum leads on logical qubit demonstrations. IonQ is the most liquid public play. Best near-term gate fidelity makes trapped ion a leading candidate for early fault tolerance.
Neutral Atoms
DevelopingThe fastest-improving modality. 1000+ atom arrays demonstrated. Operates at room temperature.
Individual neutral atoms are trapped by focused laser beams (optical tweezers) and arranged in programmable 2D or 3D arrays. Qubits interact via Rydberg states when lasers excite them to high energy levels.
Pros
- +Highly scalable (1000+ atoms demonstrated)
- +Reconfigurable mid-circuit connectivity
- +Long coherence times
- +Trap operates near room temperature
Cons
- −Slower than superconducting
- −Gate fidelities still catching up
- −Complex multi-laser optical systems
Key Companies
Investor note: The fastest-improving modality over the last 18 months. Pasqal and QuEra are the commercial leaders. planqc is the European dark horse, backed by Munich Quantum Valley.
Photonic
EarlyHigh-risk, potentially transformational. PsiQuantum is betting $2B+ that photonic chips can be manufactured at semiconductor scale.
Quantum information is encoded in particles of light (photons). Computation happens through optical interference on chip-scale waveguides. Unlike other approaches, photons can operate at room temperature and traverse fiber optic networks.
Pros
- +Room temperature operation (no cryogenics)
- +Naturally suited for quantum networking
- +Compatible with semiconductor fab processes
- +Photons travel at the speed of light
Cons
- −Photons don't interact easily — hard two-qubit gates
- −Requires extremely low-loss components
- −Error rates still high
- −Deterministic photon sources remain a challenge
Key Companies
Investor note: PsiQuantum is the biggest single bet in quantum computing — $2.1B raised on the thesis that semiconductor fabs (GlobalFoundries) can manufacture photonic quantum computers at scale. High risk, potentially category-defining if the bet lands.
Silicon Spin Qubits
EarlyThe long-term scalability bet. If it works, the semiconductor industry's manufacturing muscle wins.
Electron or nuclear spins of individual atoms in silicon act as qubits. Closely related to conventional transistor physics — qubits can in principle be manufactured using existing CMOS processes at atomic scale.
Pros
- +Potentially manufacturable at fab scale
- +Smallest physical footprint of any modality
- +Long coherence in isotopically pure silicon (Si-28)
- +Intel and TSMC manufacturing expertise
Cons
- −Least mature of major approaches
- −Requires atomic-level fabrication precision
- −Cryogenic cooling still required (~1K)
- −Control electronics scaling is unsolved
Investor note: The multi-decade scalability bet. If silicon spin qubits work, conventional chip fabs become quantum fabs. Intel's sustained investment (despite slow progress) signals strategic intent beyond near-term returns.
Quantum Annealing
Most MatureSpecialized, not universal. Already commercially deployed. The only approach with real quantum revenue today.
Uses quantum tunneling to find low-energy configurations of optimization problems — effectively letting the system "tunnel" through energy barriers that would trap classical algorithms. Not a gate-based universal quantum computer.
Pros
- +Already commercially deployed today
- +Largest qubit counts (5,000+)
- +Works for specific optimization problems now
- +Hybrid classical-quantum solvers in production
Cons
- −Only solves certain optimization problem types
- −Not universal — cannot run arbitrary algorithms
- −Results versus classical alternatives are debated
- −Limited addressable market
Key Companies
Investor note: D-Wave is the only quantum computing company with material recurring revenue today. Niche but real — logistics, scheduling, and financial optimization customers. Watch hybrid solver adoption as the near-term revenue driver.
How to Measure Progress — Benchmarking
The metrics that matter, and why qubit count alone is misleading.
Qubit count
The most cited but least meaningful metric. More qubits do not equal a better computer if they're noisy. A machine with 10 high-quality qubits may outperform one with 1,000 low-quality ones. Quantity has a quality of its own — until it doesn't.
Gate fidelity
The probability that a quantum operation gives the correct result. Two-qubit gate fidelity of 99.9% sounds high, but errors compound: 100 sequential operations yield roughly a 90% chance of a correct result. This compounding is why error correction is essential.
Coherence time
How long a qubit maintains its quantum state before decoherence destroys it. Determines how many operations you can run before the answer becomes noise. Trapped ion systems (seconds to minutes) have a large advantage over superconducting (microseconds).
Quantum Volume (QV)
IBM's metric combining qubit count, connectivity, and fidelity into a single number. Better than raw qubit count but still limited — it captures circuit complexity on a square circuit, not application performance.
CLOPS
Circuit Layer Operations Per Second — measures how many quantum circuits a system can execute per second. More practically relevant for real applications than QV, since throughput matters for iterative algorithms.
Logical qubits
Error-corrected qubits built from many physical qubits. The real milestone. IBM demonstrated 48 logical qubits in late 2023; Quantinuum demonstrated 56 in 2024. A machine with 50 logical qubits may be more practically useful than one with 1,000 noisy physical qubits.
| Modality | 2Q Gate Fidelity | Coherence Time | Qubit Count | Logical Qubits |
|---|---|---|---|---|
| Superconducting | 99.5% | ~100 µs | 1,000+ | ~48 (IBM) |
| Trapped Ion | 99.9%+ | seconds–minutes | 50–100 | 56 (Quantinuum) |
| Neutral Atom | 99.5% | seconds | 1,000+ | Early stage |
| Photonic | 95–99% | variable | variable | Research stage |
| Silicon Spin | 99%+ | seconds (Si-28) | <50 | Research stage |
Approximate figures as of early 2025. The field moves quickly.
The Road Ahead — What to Watch
Three eras, five questions, and the picks-and-shovels play.
The three eras
Now
NISQ Era
Noisy Intermediate-Scale Quantum. Hundreds of noisy qubits. Limited real-world advantage. Where we are today. Useful for research and early experiments, not production deployments.
2027–2032 est.
Early Fault-Tolerant
Logical qubits with error correction at meaningful scale. First genuine quantum advantage in specific domains — chemistry simulation, constrained optimization.
2030s+
Fault-Tolerant
Millions of physical qubits, thousands of logical qubits. Broad commercial quantum advantage. Cryptographically relevant. The full realization of the quantum promise.
Key milestones to watch
- →First demonstration of quantum advantage on a commercially relevant problem (not a synthetic benchmark)
- →100+ logical qubits with error rates below fault-tolerance threshold
- →First quantum computer integrated into a production HPC workflow
- →Post-quantum cryptography fully deployed across critical infrastructure (NIST PQC standards finalized 2024 — migration underway)
Five questions for evaluating a quantum company
Physical qubits or logical qubits — and what is the error rate?
Physical qubit counts are marketing. Logical qubit counts with sub-threshold error rates are the real measure of progress.
What is the path to fault tolerance — and how many physical qubits does it require?
Surface codes require ~1,000 physical qubits per logical qubit. Some companies claim novel codes that need far fewer. Scrutinize the assumptions.
Who are the paying customers today — not just pilot partners?
Many quantum companies have long lists of "partners" that are unpaid research relationships. Recurring revenue is the only honest signal of product-market fit.
What happens to this company if a rival modality wins?
Pure-play hardware companies face existential modality risk. Software and tooling companies (algorithms, compilers, orchestration) are more modality-agnostic.
Is the moat in hardware, software, or both?
IBM's real moat is Qiskit and 500,000+ registered users. Google's is Cirq and research talent. Software lock-in may prove more durable than hardware architecture.