Investor's Guide
Last updated: May 2026
Understanding Quantum Computing
A plain-language guide to the science, the hardware race, the investment landscape, and what it all means for investors and professionals navigating the quantum era.
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.
Why Classical Computers Have Limits
For 60 years, Moore's Law held: the number of transistors on a chip doubled roughly every two years, reliably delivering faster and cheaper computers. That era is ending. Transistors are now approaching atomic scale — modern chips pack features just a few atoms wide. You cannot make a transistor smaller than a single atom.
The more fundamental problem is computational complexity. Some problems scale exponentially with input size, meaning classical computers become impractical not just slow. Consider a 300-variable optimization problem — the number of possible configurations exceeds the number of atoms in the observable universe (roughly 1080). No amount of Moore's Law can solve that. Quantum computers are engineered specifically for this class of problem.
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.
The Qubit in Detail
The differences between classical bits and qubits run deeper than just “superposition.” Every property has engineering consequences:
| Property | Classical Bit | Qubit |
|---|---|---|
| States | 0 or 1 | Superposition of 0 and 1 |
| Operations | Logic gates | Quantum gates |
| Error rate | ~0% | 0.1–5% (today) |
| Temperature | Room temp | ~15 millikelvin |
| Copying | Easy | Impossible (no-cloning theorem) |
The No-Cloning Theorem
One of the strangest properties of quantum mechanics: you cannot make an exact copy of an unknown quantum state. This sounds like a limitation — and it is, for some applications. But it's also profoundly useful for security. In quantum key distribution (QKD), if an eavesdropper tries to intercept a quantum message, the act of measurement disturbs the quantum state. The disturbance is detectable. Any eavesdropper reveals themselves. This is the foundation of quantum cryptography — not just theoretically secure, but provably so under the laws of physics.
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.
Quantum vs Classical Parallelism — The Misconception
The most widespread misunderstanding about quantum computing: “it tries all answers at once.” If this were true, you could just read out all 2n answers simultaneously. You can't. Measurement collapses the superposition to a single result. The real power is interference. Quantum algorithms are carefully designed so that wrong answers cancel each other out (destructive interference) while correct answers reinforce each other (constructive interference) — exactly like noise-canceling headphones, which use destructive interference to eliminate background noise. The art of quantum algorithm design is engineering this interference correctly.
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.
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 — with the numbers that matter.
Drug Discovery — The $1.5 Trillion Opportunity
Classical computers can accurately simulate molecules up to about 50 atoms. Beyond that, the quantum mechanical interactions become too complex — the compute cost scales exponentially with molecule size. Drug molecules typically contain hundreds to thousands of atoms. The result: pharmaceutical R&D relies heavily on approximations and brute-force screening.
Quantum computers simulate quantum systems naturally. A fault-tolerant quantum computer could model proteins with full quantum mechanical accuracy, dramatically improving hit rates in early-stage drug discovery. Pfizer's internal estimates suggest quantum computing could compress drug discovery timelines from 12 years to 3–5 years. The total addressable market: a $1.5 trillion global pharmaceutical industry where the average drug costs $2.6B to bring to market.
Optimization at Scale — Logistics, Finance, Energy
The Traveling Salesman Problem with just 50 cities has more possible routes than there are atoms in the universe. Airlines, logistics companies, financial institutions, and power grid operators face structurally similar problems every day. Classical heuristics find good-enough solutions. Quantum optimization algorithms, when mature, could find genuinely optimal or near-optimal solutions.
FedEx has estimated that quantum optimization of its routing alone could save $150M per year in fuel costs. JPMorgan Chase and Goldman Sachs both have quantum research teams focused on portfolio optimization and risk analysis. Volkswagen ran a quantum-optimized traffic routing pilot in Lisbon in 2019 — the first real-world deployment of quantum optimization.
Battery and Materials Science
Next-generation batteries require understanding the quantum chemistry of lithium-ion and lithium-sulfur cells at the atomic level. Current classical simulations use approximations that make it impossible to fully predict how new material compositions will behave. BMW, BASF, and Volkswagen are already running quantum chemistry experiments, targeting breakthroughs in energy density and cycle life. The electric vehicle transition makes this one of the most commercially urgent application areas.
Cryptography — Breaking Encryption
Shor's algorithm can break RSA encryption on a sufficiently large fault-tolerant quantum computer. This is why governments worldwide are funding post-quantum cryptography. NIST finalized its first four PQC standards in 2024 — migration has already begun across critical infrastructure.
Nation-state actors are collecting encrypted government, financial, and military data today, storing it, and waiting for quantum computers powerful enough to decrypt it. The data is useless now but may not be in 10–15 years. The NSA issued guidance in 2022; NIST finalized four post-quantum cryptographic standards in August 2024. Any organization not actively migrating to PQC is accumulating cryptographic risk that compounds with every day of delay. Financial institutions, healthcare systems, and defense contractors should treat PQC migration as a board-level risk item, not an IT project.
What quantum will not do
The Investment Landscape
$40 billion in, with the hard part still ahead. How to think about quantum as an investment.
The $40 Billion Question
Total venture and government investment in quantum computing exceeded $40 billion by end of 2025. This is not vaporware — it represents a genuine geopolitical and industrial race. The funding breakdown:
China
$15B+
Nat'l strategy
EU
$7.2B
Quantum Flagship
Japan
$7.4B
National plan
US
$2.5B
DOE + CHIPS Act
UK
$2.5B
10-year strategy
Private VC
$3.77B
2025 alone
Private VC grew 58% year-over-year in 2025. Government funding reflects national security imperatives as much as commercial ones.
Three Ways to Invest in Quantum Today
1. Pure-Play Public Companies
IonQ (IONQ), Rigetti (RGTI), D-Wave (QBTS), Quantum Computing Inc. (QUBT), Arqit (ARQQ) — directly exposed to quantum's commercialization timeline. High volatility. Potential multi-bagger returns if quantum advantage materializes. Also potential zero if timelines extend another decade.
Best for: investors who have done deep technical diligence and are comfortable with binary outcomes over a 5–10 year horizon.
2. Quantum ETFs
QTUM (Defiance), WQTM (WisdomTree), CHPX (Global X) — diversified exposure across pure-plays and enabling technology. Lower volatility than individual names. Includes enabling tech companies (classical semiconductor firms) as downside protection.
Best for: investors wanting quantum exposure without single-company technical risk. Check expense ratios and holdings composition — some ETFs are more “pure quantum” than others.
3. Large-Cap Indirect Exposure
IBM, Google (Alphabet), NVIDIA, Honeywell (Quantinuum parent), Intel — quantum upside optionality with classical computing stability as a floor. IBM generates ~$600M/year from quantum-related services. NVIDIA's CUDA-Q platform positions it as the operating system of hybrid quantum-classical computing.
Best for: investors who believe quantum will matter but want downside protection from world-class classical businesses.
Valuation Frameworks
Traditional financial metrics fail for most quantum companies. Revenue multiples are meaningless for pre-revenue companies (most pure-plays). Better frameworks:
- →Technical milestone progression: Is the company hitting its published roadmap milestones? Slippage is a yellow flag; repeated slippage is a red flag. Compare milestones against peer companies.
- →Talent density: Quantum is a talent-constrained industry. Count PhD physicists, particularly those who have published in top journals (Nature, Science, Physical Review Letters). Talent concentrations predict technical capability.
- →Partnership quality: Distinguish between paid commercial partnerships (revenue signal) and research MoUs (marketing signal). IBM's 400+ paying enterprise clients is a fundamentally different signal than an unpaid research collaboration.
- →IP portfolio and publication record: Strong companies publish openly and have defensible patent portfolios. Secrecy combined with bold claims without peer review is a major red flag.
- →Path to fault tolerance: What error correction code? How many physical qubits per logical qubit? What is the required fidelity threshold? Companies that cannot answer these questions concretely are speculating about their own roadmap.
Red flags: companies claiming quantum advantage without peer review, vague timelines, hardware claims without published gate fidelity data, and excessive focus on physical qubit counts.
The Timeline Risk — Be Honest
Every major quantum roadmap has slipped. IBM's 2023 roadmap was revised. Google's supremacy claim (2019) was contested. The “useful quantum computers in 10 years” prediction has been made continuously since 1994. This is not a reason to dismiss the field — the underlying physics is real and progress is genuine. It is a reason to stress-test your timeline assumptions.
The honest investor framework: what does this company's technology look like if fault tolerance arrives in 5 years? In 10 years? In 20 years? Which companies survive all three scenarios?
The Hardware Race — Seven Approaches
The quantum industry is not converging on a single technology. Seven distinct physical implementations are in active competition, each with different strengths, timelines, and investment profiles.
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 (1,000+)
Cons
- −Requires extreme cooling — bulky, expensive cryostats
- −Short coherence times (microseconds)
- −Scaling wiring is a major bottleneck
Best Published Result
Google Willow (Dec 2024): demonstrated below-threshold error correction, a key milestone. IBM Heron: 99.5% two-qubit gate fidelity on a 133-qubit chip.
Key Challenge
Routing thousands of control wires from room temperature into the dilution refrigerator without introducing noise — the “wiring problem” becomes exponentially harder as qubit counts scale.
Timeline
5–8 years to early fault-tolerant demonstrations at commercially useful scale; already the most cloud-accessible modality.
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. Rigetti is the most liquid pure-play superconducting stock.
Trapped Ion
Most MatureCompeting with superconducting for leadership. Best gate fidelity of any modality today. Quantinuum leads the logical qubit race.
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, eliminating manufacturing variation.
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
Best Published Result
Quantinuum H2-1 (2025): 48 logical qubits demonstrated with error correction; 99.9%+ two-qubit gate fidelity. IonQ Forte: 36 algorithmic qubits (IonQ's proprietary benchmark).
Key Challenge
Scaling from tens to hundreds of ions requires new trap architectures — “shuttling” ions between trap zones introduces errors and latency that don't exist in small systems.
Timeline
3–6 years to meaningful fault-tolerant operations; currently the modality closest to near-term commercial quantum advantage on chemistry problems.
Key Companies
Investor note: Quantinuum leads on logical qubit demonstrations and is majority-owned by Honeywell — a strategic investor with deep pockets and patience. IonQ is the most liquid public play. Best near-term gate fidelity makes trapped ion a leading candidate for early fault tolerance demonstrations.
Neutral Atoms
DevelopingThe fastest-improving modality. 1,000+ atom arrays demonstrated. Room-temperature trapping with a reconfigurable architecture that superconducting can't match.
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. The trap structure can be reconfigured mid-circuit.
Pros
- +Highly scalable (1,000+ atoms demonstrated)
- +Reconfigurable mid-circuit connectivity
- +Long coherence times
- +Trap operates near room temperature
Cons
- −Slower than superconducting
- −Gate fidelities still catching up to trapped ion
- −Complex multi-laser optical systems
Best Published Result
QuEra/Harvard collaboration (2024): 48 error-corrected logical qubits, 228 physical qubits. Pasqal: 100+ atom arrays with 99.5% two-qubit gate fidelity achieved in 2024.
Key Challenge
Maintaining atom positioning precision and gate fidelity across arrays of 1,000+ atoms — any vibration or laser instability affects every qubit simultaneously.
Timeline
5–10 years; improving fastest of any modality over the past 24 months. The 2024 QuEra logical qubit result was a genuine breakthrough that accelerated investor interest.
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. Atom Computing was acquired by Microsoft in 2023, giving Microsoft a neutral-atom hedge alongside its topological bet.
Photonic
EarlyHigh-risk, potentially transformational. PsiQuantum is betting $2B+ that photonic chips can be manufactured at semiconductor scale. The only modality that can run at room temperature.
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 natively.
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 vs other modalities
- −Deterministic photon sources remain an unsolved challenge
Best Published Result
Xanadu Borealis (2022): quantum computational advantage on Gaussian boson sampling (Nature, peer-reviewed). PsiQuantum: no public qubit metrics published yet.
Key Challenge
Building deterministic, high-efficiency single-photon sources — current probabilistic sources require massive overhead that scales unfavorably for fault-tolerant architectures.
Timeline
8–12+ years to commercial scale. PsiQuantum's manufacturing-first bet is a binary outcome; if GlobalFoundries can yield photonic chips at scale, the timeline compresses dramatically. No qubit results published makes independent validation impossible today.
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. Their silence on qubit metrics is either disciplined stealth or a signal that no results exist yet. High risk, potentially category-defining if the fab bet lands.
Silicon Spin Qubits
EarlyThe long-term scalability bet. If it works, the semiconductor industry's manufacturing muscle wins. Still years from demonstrating multi-qubit error correction.
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 potentially applicable
Cons
- −Least mature of major approaches for multi-qubit systems
- −Requires atomic-level fabrication precision
- −Cryogenic cooling still required (~1K)
- −Control electronics scaling is unsolved
Best Published Result
Intel Tunnel Falls (2023): 12-qubit chip shipped to research partners. Diraq (2023): 99.9% single-qubit gate fidelity in silicon, published in Nature.
Key Challenge
Achieving consistent two-qubit gate fidelity across manufactured chips — single-atom defects that are irrelevant in classical chips are fatal flaws in silicon spin qubit devices.
Timeline
10–15 years to commercial scale; the most distant but potentially most scalable. Intel's continued investment signals long-term conviction despite slow near-term progress.
Investor note: The multi-decade scalability bet. If silicon spin qubits work, conventional chip fabs become quantum fabs overnight. Intel's sustained investment (despite slow near-term progress) signals strategic intent beyond near-term returns. Quantum Motion's CMOS-compatible approach is particularly interesting for fab partnerships.
Topological Qubits (Microsoft)
Very EarlyMicrosoft's moonshot. Uses exotic quantum states of matter to create inherently error-protected qubits. In February 2025, Microsoft unveiled Majorana 1 — the first topological qubit processor. If it works at scale, it could leapfrog every other approach.
Uses exotic quantum states of matter called Majorana fermions — particles that are their own antiparticle — to encode quantum information in a fundamentally protected way. The topology of the qubit state means certain types of errors cannot occur, potentially requiring far fewer physical qubits per logical qubit than conventional approaches.
Pros
- +Inherently protected against certain error types
- +Could require dramatically fewer physical qubits for fault tolerance
- +Microsoft's deep pockets sustain long research timelines
- +Azure integration provides instant commercial distribution if successful
Cons
- −Majorana particles were disputed for years; evidence is still contested
- −No gate fidelity results published yet
- −Technical claims have faced peer review challenges
- −Still unproven at any computational scale
Best Published Result
Microsoft Majorana 1 (Feb 2025): first topological qubit chip announced. No gate operation fidelity metrics published. A Nature paper was retracted in 2021 due to data presentation issues — Microsoft has since rebuilt its experimental program.
Key Challenge
Proving that Majorana particles can be reliably created, controlled, and braided to perform computation — the underlying physics is still being established at the level of single-qubit operations.
Timeline
Highly uncertain. Microsoft claims a path to a million-qubit system within a decade, enabled by the reduced error correction overhead. Skeptics believe the physics challenges could set the timeline back significantly. Follow Nature publications closely.
Key Companies
Investor note: Binary bet on one of the most audacious physics gambles in computing history. If topological qubits work as theorized, Microsoft could leapfrog IBM and Google in a single generation. If the physics proves intractable, the program terminates. There is no middle outcome. For investors, the exposure is through Microsoft stock — topological is a call option embedded in a $3T company.
Quantum Annealing
Most MatureSpecialized, not universal. Already commercially deployed at scale. D-Wave is the only quantum computing company generating significant recurring revenue today — a fact often overlooked by investors chasing gate-based stories.
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. Problems must be formulated as Quadratic Unconstrained Binary Optimization (QUBO) problems.
Pros
- +Already commercially deployed today
- +Largest qubit counts (7,000+ in Advantage2)
- +Works for specific optimization problems now
- +Hybrid classical-quantum solvers in production use
- +~$60M ARR — real revenue, real customers
Cons
- −Only solves optimization problems expressible in QUBO form
- −Not universal — cannot run Shor's or Grover's algorithms
- −Performance vs classical alternatives is problem-dependent
- −Limited addressable market vs gate-based universal quantum
Best Published Result
D-Wave Advantage2 (2023): 7,000+ qubit processor. FY2024 ARR of approximately $60M with customers including Volkswagen, Mastercard, JDSU, and government agencies. First quantum company with publicly reported recurring commercial revenue.
Key Challenge
Demonstrating quantum speedup over classical heuristics on real-world optimization problems at a scale that justifies the hardware cost — the academic debate on quantum annealing vs simulated annealing remains unresolved.
Timeline
Already deployed commercially. Revenue growing ~25% YoY. The relevant question is not “when will it be useful” but “how large can the optimization market be for this specific approach.”
Key Companies
Investor note: D-Wave is the only quantum computing company with material recurring revenue today. Niche but real — logistics, scheduling, financial optimization, and drug discovery customers. The stock trades at a revenue multiple, unlike peers that trade on speculation. Watch hybrid solver adoption (classical + quantum annealing) as the near-term revenue driver. Undervalued by investors who dismiss it as “not real quantum.”
How to Measure Progress — Benchmarking
The metrics that matter, why qubit count alone is misleading, and how to read a quantum research paper.
Why Comparing Quantum Computers Is Hard
Each major company uses different metrics, often designed to make their particular system look best. IBM uses Quantum Volume. Google uses XEB (cross-entropy benchmarking). IonQ invented Algorithmic Qubits. These metrics are not directly comparable — by design. When a company announces a benchmark record, the first question to ask is: who defined the benchmark, and did rivals agree to use it?
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. Qubit count has marketing value; it has limited technical value without fidelity data.
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 for anything beyond small circuits.
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). Coherence time divided by gate time gives the maximum useful circuit depth.
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. IonQ scored QV of 32,768 in 2022, demonstrating trapped ion's fidelity advantage.
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 (VQE, QAOA). Superconducting systems dominate on CLOPS despite lower fidelity.
Logical qubits
Error-corrected qubits built from many physical qubits. The real milestone that separates NISQ from fault-tolerant quantum computing. Quantinuum demonstrated 48 logical qubits in 2025; QuEra/Harvard achieved 48 logical qubits using neutral atoms in 2024. A machine with 50 logical qubits may be more practically useful than one with 1,000 noisy physical qubits.
The Logical Qubit Race — Where Things Stand
Logical qubits are the metric that actually matters for fault-tolerant computing. Current leaders as of 2025:
Trapped ion; highest quality logical qubits demonstrated to date
Neutral atoms; breakthrough result published in Nature
First superconducting system to demonstrate error rates improving with scale
Focus on scalable code architecture over maximum logical qubit count
Reading a Quantum Research Paper — Investor's Quick Guide
You don't need a physics PhD to evaluate quantum research. Look for:
Green flags
- + Published in Nature, Science, or PRL
- + Explicit comparison to classical simulation
- + Reports physical AND logical qubit metrics
- + Gate fidelity measured at circuit depth, not single-gate
- + Independent replication or third-party verification
Red flags
- − “Quantum advantage” on a synthetic benchmark
- − No classical comparison baseline provided
- − Proprietary metrics with no standard equivalent
- − Peak fidelity on one qubit, not system average
- − Press release without preprint or paper link
| Modality | 2Q Gate Fidelity | Coherence Time | Qubit Count | Logical Qubits |
|---|---|---|---|---|
| Superconducting | 99.5% | ~100 µs | 1,000+ | ~12 (IBM, 2024) |
| Trapped Ion | 99.9%+ | seconds–minutes | 50–100 | 48 (Quantinuum, 2025) |
| Neutral Atom | 99.5% | seconds | 1,000+ | 48 (QuEra, 2024) |
| Photonic | 95–99% | variable | variable | Research stage |
| Silicon Spin | 99%+ (1Q) | seconds (Si-28) | <20 | Research stage |
| Topological | N/A | N/A | 1 chip | Not yet demonstrated |
| Annealing | N/A | microseconds | 7,000+ | N/A (not gate-based) |
Approximate figures as of early 2025. The field moves quickly — always check publication dates on benchmark claims.
The Road Ahead — What to Watch
Three eras, five scenarios, the quantum internet, 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. Early commercial deployments.
2030s+
Fault-Tolerant
Millions of physical qubits, thousands of logical qubits. Broad commercial quantum advantage. Cryptographically relevant. The full realization of the quantum promise.
What Fault Tolerance Actually Requires
Error correction is the central unsolved engineering problem of quantum computing — and the numbers are sobering. Surface codes (the leading error correction approach) require approximately 1,000 physical qubits per logical qubit. For drug discovery applications you need around 1 million logical qubits. That's approximately 1 billion physical qubits. The current physical qubit record is around 1,000 qubits.
The gap is enormous. But so is the progress rate. In 2019, Google demonstrated quantum supremacy on 53 qubits. In 2024, they demonstrated below-threshold error correction. Each generation of chips roughly doubles qubit count and improves fidelity. Novel error correction codes (LDPC codes, bivariate bicycle codes) may reduce the overhead significantly — IBM's Gross code reduces the overhead by roughly 10x compared to surface codes.
The Quantum Internet
Quantum networking — connecting quantum computers over fiber using entangled photons — is already moving from research to early commercial deployment. Quantum Key Distribution (QKD) networks are operational today: QuantumCTek's Beijing-Shanghai quantum backbone (2,000km), BT and Nu Quantum's UK trials, and DARPA-funded US metro networks.
The long-term vision: a global quantum internet where quantum processors communicate via entangled links, enabling distributed quantum computing, unhackable communications, and quantum-secured financial systems. Near-term investment angles: QKD hardware providers (QuantumCTek, Toshiba, ID Quantique), quantum repeater research, and satellite-based quantum communication (China's Micius satellite has demonstrated intercontinental QKD).
Five Scenarios for 2035
Portfolio construction depends on which scenario you assign the highest probability. These are not equally likely:
Quantum Winter
Low (~10%)Decoherence proves harder than expected. No fault tolerance by 2035. Pure-play hardware companies consolidate or fail. Enabling tech companies (cryogenics, control electronics, software) survive. A replay of AI winters: the technology is real, just not yet.
Slow Burn
Moderate (~25%)Fault tolerance arrives around 2035 in limited domains. First applications in chemistry simulation and constrained optimization. Stock prices reflect early commercial revenue from 1–2 use cases. Most pure-plays have pivoted to hybrid classical-quantum services.
Base Case
Most likely (~40%)Fault-tolerant quantum computers demonstrated by 2030–2032, commercial deployment 2033–2035 in chemistry, materials, and financial optimization. Two to three hardware platforms win. Software and algorithm companies deliver the first wave of enterprise value.
Accelerated
Possible (~20%)One modality achieves a breakthrough in error correction by 2027–2028, triggering massive consolidation. Venture and corporate investment surges. The winning company becomes the AWS of quantum computing. Competing approaches lose funding rapidly.
Quantum Surprise
Unlikely but possible (~5%)A new modality or algorithmic approach emerges that wasn't on anyone's roadmap — analogous to how transformer architectures reshaped AI. Incumbent hardware advantages evaporate. The most likely candidate: topological or some combination of modalities not yet named.
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 — the true capability milestone
- →First quantum computer integrated into a production HPC or cloud workflow at scale
- →Post-quantum cryptography fully deployed across critical infrastructure (NIST PQC standards finalized 2024 — migration underway, but slow)
- →First quantum networking demonstration that connects two physically separate quantum processors in a useful computation
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. A company that only reports physical qubits is either still in NISQ territory or deliberately obscuring its logical qubit 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. Ask: has this been demonstrated experimentally, or is it a theoretical claim?
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. D-Wave is the benchmark: ~$60M ARR from real enterprise customers.
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. Ask: does this company's software run on IBM and Quantinuum hardware, or only their own?
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. Quantinuum's is TKET and enterprise chemistry customers. Software lock-in may prove more durable than hardware architecture advantages that could be leap-frogged by a new modality.
Key Terms — Quantum Computing Glossary
20 essential terms defined in plain English. No PhD required.
The basic unit of quantum information. Unlike a classical bit (0 or 1), a qubit can exist in a superposition of both states until measured. The physical implementation varies by hardware modality: superconducting circuits, trapped ions, photons, etc.
A quantum mechanical state in which a qubit exists as a combination of 0 and 1 simultaneously. When measured, it collapses to a definite state. Not magic — it's a well-understood probabilistic state described by quantum mechanics.
A quantum correlation between two or more qubits such that measuring one instantly determines the other's state, regardless of distance. A computational resource that allows quantum computers to process correlated information in ways classical computers cannot.
The mechanism by which quantum algorithms actually work. Correct computational paths are amplified (constructive interference); incorrect paths are cancelled (destructive interference). Not brute-force parallelism — careful algorithm design is required.
The process by which a qubit loses its quantum properties through interaction with its environment. Heat, vibration, and electromagnetic noise all cause decoherence. The primary engineering challenge in quantum hardware. Minimized by extreme cooling and isolation.
Noisy Intermediate-Scale Quantum. Coined by physicist John Preskill in 2018. Describes current quantum computers: 50–1,000 qubits, too noisy for error correction, limited circuit depth. Most quantum computers today are NISQ devices.
The probability that a quantum gate operation produces the correct result. Expressed as a percentage. Two-qubit gate fidelity of 99.9% means 0.1% error per operation. Errors compound with circuit depth, which is why 99.9% is very different from 99%.
How long a qubit maintains its quantum state before decoherence destroys the computation. T1 (relaxation time) and T2 (dephasing time) are the two key measures. Trapped ions: seconds to minutes. Superconducting: microseconds.
A class of techniques that encode quantum information redundantly across multiple physical qubits so errors can be detected and corrected without measuring (and collapsing) the quantum state. Essential for fault-tolerant computing.
The leading quantum error correction code. Encodes one logical qubit in a 2D grid of physical qubits. Requires approximately 1,000 physical qubits per logical qubit for typical target error rates. Favored for its local operations (neighboring qubit interactions only).
An actual hardware qubit — a superconducting circuit, a trapped ion, a photon. Subject to noise and errors. Contrasted with logical qubits. Current record: ~1,000 superconducting qubits (IBM), 7,000+ annealing qubits (D-Wave).
An error-corrected qubit built from many physical qubits using quantum error correction. The unit of computation in fault-tolerant quantum computing. A machine with 50 logical qubits may outperform one with 1,000 noisy physical qubits for real algorithms.
A quantum computer that can perform arbitrarily long computations by continuously correcting errors. Requires logical qubits with error rates below the fault-tolerance threshold (~10⁻¹⁰ per operation). The holy grail of the field.
A hardware-agnostic benchmark developed by IBM that measures the largest random circuit a quantum computer can run with >⅔ probability of success. Incorporates qubit count, connectivity, and gate fidelity. Higher is better.
The point at which a quantum computer solves a problem faster or more accurately than the best classical computer. Sometimes called quantum supremacy. Contested: requires careful specification of which problem, which classical algorithm, and what hardware.
A quantum algorithm that can factor large integers exponentially faster than any known classical algorithm. Breaks RSA and ECC encryption, the foundation of internet security. Requires millions of fault-tolerant logical qubits. Not runnable on any current hardware.
A quantum algorithm providing a quadratic speedup for unstructured search problems. Less dramatic than Shor's but more broadly applicable. Gives a quantum computer a theoretical advantage over classical brute-force search — relevant for optimization and database problems.
Classical (non-quantum) cryptographic algorithms designed to resist attacks from quantum computers running Shor's algorithm. NIST finalized four PQC standards in 2024 (CRYSTALS-Kyber, CRYSTALS-Dilithium, FALCON, SPHINCS+). Migration is underway across critical infrastructure.
A quantum optimization technique that uses quantum tunneling to find low-energy solutions to combinatorial optimization problems. Not a universal quantum computer — cannot run Shor's algorithm. D-Wave is the primary commercial vendor. Already deployed commercially.
A quantum communication method that uses quantum mechanics to distribute cryptographic keys with information-theoretic security. Any eavesdropping disturbs the quantum state and is detectable. Already deployed commercially (China, UK, EU). Different from — and complementary to — post-quantum cryptography.