Quantum Computing
What it is, why it is fundamentally different, how close it really is, and why it threatens the encryption that protects everything.
A quantum computer is not a faster computer; it is a different kind of computer that can solve a few specific problems no classical machine ever will. It will not replace your laptop, but for the problems it fits, it changes what is possible.
What this argues
- Quantum is not 'faster classical'It is good at a narrow set of problems, useless for most.
- The hard part is errorsQubits are fragile; error correction is the whole game.
- Real advantage is years awayMeaningful machines are a 2030s story; IBM, Quantinuum and others target fault tolerance by 2028-2029.
- But the threat is now'Harvest now, decrypt later' means encrypted data is already at risk.
- Value in chemistry and materialsSimulating nature is where quantum should first pay off.
What has changed
Updated November 2026
Money surged and quantum companies are going public
- McKinsey's Quantum Technology Monitor (20 Apr 2026) puts 2025 investment at $12.6B, 6.3 times 2024, and 2025 quantum computing revenue above $1B. It projects a $60-100B quantum technology market by 2035 ($43-71B for computing). A commentator calls the investment figure directionally correct but imprecise.
- Quantinuum's amended S-1A (28 May 2026, via Quantum Computing Report) set IPO pricing at $45-50 per share, implying $11.43-12.70B, plus a $100M CHIPS Act award.
- Other deals reported by Quantum Computing Report and The Quantum Insider (Jun 2026): Atom Computing raised over $300M, Pasqal agreed a Nasdaq merger at a $2.0B valuation, Terra Quantum a SPAC at $3.5B, and Infleqtion listed on the NYSE after raising about $550M.
Breaking RSA-2048 now needs under a million qubits
Google researcher Craig Gidney's May 2025 paper estimates a machine with fewer than one million noisy qubits could factor RSA-2048 in about five days, versus roughly 20 million in 2019 (The Quantum Insider, 24 May 2025). A June 2026 summary on postquantum.com reports a February 2026 vendor claim (Iceberg Quantum) of under 100,000 qubits using LDPC codes; this is a single vendor claim.
Google reports a verifiable speed-up claim
Google's Quantum Echoes algorithm was reported to run 13,000 times faster than the best classical algorithm (The Quantum Insider, Jun 2026). Willow remains at 105 qubits. This is a narrow demonstration, not a fault-tolerant machine.
IBM and Quantinuum hardware and roadmap moves
- IBM's Nighthawk (120 qubits, square lattice) was validated in two independent studies (Quantum Computing Report, 20 Jun 2026). IBM's technology page still targets Starling in 2029: 200 logical qubits and 100 million gates.
- IBM is investing over $10B in quantum toward a fault-tolerant system by 2029, with a planned foundry drawing a $1B CHIPS Act incentive (Quantum Computing Report, 29 May 2026).
- Quantinuum's Helios (98 qubits) was validated by Sandia and is on cloud access, with a 2:1 physical-to-logical ratio (Quantum Computing Report, 18 Jun 2026).
- Microsoft's Majorana 2 reported parity lifetimes improving from milliseconds to over 20 seconds (The Quantum Insider, Jun 2026).
Governments pull fault-tolerance dates forward
- The US DOE's Quantum Genesis targets a fault-tolerant computer by 2028 with 150-250 logical qubits (Quantum Computing Report, 24 Jun 2026).
- Executive orders directed federal agencies to transition to post-quantum cryptography (Quantum Computing Report, 22 Jun 2026).
- QuEra plans a system targeting 256 logical qubits from over 10,000 physical qubits by 2028; Pasqal targets 200+ logical qubits by 2029 (Quantum Computing Report and The Quantum Insider, Jun 2026).
NIST finalises its migration timeline report
NIST's page for the IR 8547 draft states it was obsoleted on 27 Oct 2026 by the final IR 8547, covering quantum-vulnerable standards and their replacements (NIST CSRC). The final text was not read, so the deck's 2030 and 2035 dates are unchanged.
Not refreshed: the 22-bit largest-key-cracked figure, the market forecast chart (MarketsandMarkets), the public funding by country chart, the qubit-count chart, and the 2030/2035 RSA deprecation dates (final NIST IR 8547 text not read).
What Quantum Is
The strange part
A classical bit is a coin lying flat, heads or tails. A qubit is a coin spinning in the air, both at once, until you look.
That single difference is where all of quantum computing's power comes from.
Bit versus qubit
| Classical bit | Qubit | |
|---|---|---|
| State | 0 or 1, one at a time | 0 and 1 at once (superposition) |
| Logic | deterministic switches | probabilities and interference |
| Scaling | n bits = n values | n qubits = 2^n values at once |
| Good at | everything you use daily | a few very hard problems |
What makes qubits special
| Power | What it means |
|---|---|
| Superposition | a qubit can be 0 and 1 simultaneously, exploring many paths at once. |
| Entanglement | qubits link so that measuring one instantly informs the other. |
| Interference | wrong answers cancel out and right answers reinforce, if you design it well. |
Which problems actually suit a quantum computer
| Problem | Fit |
|---|---|
| Spreadsheets | trivially classical (classical wins) |
| Databases | classical |
| Optimisation | maybe, someday |
| Cryptography | Shor's algorithm |
| Molecules | quantum simulates quantum (quantum-only) |
More than the universe
300 qubits hold more states than there are atoms in the cosmos. 2^300 is roughly 10^90, vastly more than the ~10^80 atoms in the observable universe. That is the scale of information a few hundred perfect qubits could, in principle, manipulate.
What quantum is good, and bad, at
| Verdict | Area | Detail |
|---|---|---|
| Great | Simulating nature | molecules, materials, chemistry: quantum simulating quantum. |
| Great | Certain maths | factoring, search, some optimisation, with the right algorithm. |
| Useless | Everyday computing | email, spreadsheets, video: classical wins, always. |
| Unproven | Most 'AI + quantum' | much-hyped, little demonstrated so far. |
How It Works
From physics to a machine
How a quantum program actually runs
- Prepare: set qubits into a known starting state near absolute zero.
- Superpose: put them into combinations of 0 and 1 at once.
- Entangle and gate: apply quantum operations that interfere the possibilities.
- Measure: read out, collapsing to one answer, with high probability correct.
There is no agreed 'transistor' for qubits yet
| Approach | Players and traits |
|---|---|
| Superconducting | IBM, Google: fast, but need near-absolute-zero fridges. |
| Trapped ions | IonQ, Quantinuum: very accurate, but slower. |
| Neutral atoms | Atom Computing, QuEra: promising for scaling up. |
| Photonic | PsiQuantum, Xanadu: room-temperature, natural for networking. |
Colder than space
Superconducting qubits run at ~15 millikelvin. That is far colder than the vacuum of deep space (2.7 K). A quantum computer's dilution refrigerator is one of the coldest places in the known universe, the price of holding qubits still enough to compute.
Every qubit technology trades speed for stability
| Superconducting | Trapped ion | |
|---|---|---|
| Speed | very fast | slower |
| Accuracy | lower per gate | highest |
| Temperature | near absolute zero | room-ish, lasers |
| Scaling | wiring gets hard | connectivity limits |
| Backers | IBM, Google | IonQ, Quantinuum |
Qubit counts are rising fast, but raw count isn't the point
View the data
| Physical qubits | |
|---|---|
| 2016 | 5 |
| 2019 | 55 |
| 2021 | 140 |
| 2022 | 450 |
| 2023 | 1121 |
| 2025 | 1400 |
| 2029 | 100000 |
Source: Company disclosures; IBM roadmap · values read from the chart, approximate
The 2029 point is IBM's target of ~100,000 qubits. Values are read from a log-scale axis; 2023 is IBM's Condor at 1,121 qubits.
The Error Problem
Why it's so hard
Qubits are absurdly fragile
A stray vibration, photon or magnetic field destroys the computation. This is the only problem that matters. Solving errors, not adding qubits, is what stands between demos and useful machines.
How many fragile qubits make one reliable one
- Physical qubits: hundreds of noisy, error-prone physical qubits.
- Encode: spread one unit of information across all of them.
- Detect and fix: constantly measure for errors and correct them, live.
- Logical qubit: the result is one stable, usable 'logical' qubit.
A useful computer is a pyramid of error correction
| Layer | What it is |
|---|---|
| Useful algorithm | the actual problem you want solved |
| Logical qubits | hundreds needed: stable, error-corrected |
| Physical qubits | ~1,000 noisy qubits per logical qubit |
| Raw hardware | millions of physical qubits, near absolute zero |
We are in the noisy era, not the useful one
| NISQ (today) | Fault-tolerant (the goal) | |
|---|---|---|
| Qubits | noisy, uncorrected | error-corrected, logical |
| Scale | 10s-1,000s physical | millions physical |
| Use | experiments and demos | real problems |
| Value | learning, not advantage | genuine advantage |
Below threshold
Google's Willow crossed a 30-year milestone in 2024. For the first time, adding more physical qubits made the error rate go DOWN, not up, proving error correction can work at scale. It is the most important quantum result of the decade.
The Race
Who is building it
A tiny market today, growing fast
View the data
| Market size | |
|---|---|
| 2023 | 1.3 |
| 2025 | 3.5 |
| 2027e | 8.6 |
| 2030e | 20.2 |
Source: MarketsandMarkets (2025); wide range across forecasters
A mix of giants, specialists and nations
| Player | Approach |
|---|---|
| IBM | superconducting; the most detailed public roadmap to fault tolerance. |
| superconducting; the Willow error-correction breakthrough. | |
| IonQ / Quantinuum | trapped ions; leaders on qubit quality. |
| PsiQuantum | photonic; betting big on a million-qubit machine. |
| Microsoft / Amazon | cloud access plus exotic qubit research. |
| China | a massive, state-funded national effort. |
How the field stacks up
Source: The Quantum Insider (Jun 2026); McKinsey Quantum Technology Monitor (Apr 2026); PitchBook data via McKinsey.
The path to a useful machine is now visible
- Supremacy: Google beats a supercomputer at one task
- 1,121 qubits: IBM's Condor, largest to date
- Willow: error correction below threshold
- Fault tolerance: IBM targets 200 logical qubits
- Advantage: first genuinely useful problems
Governments are treating quantum as strategic
View the data
| Public funding | |
|---|---|
| China | 15 |
| EU + UK | 11 |
| United States | 7 |
| Japan | 3 |
| Others | 6 |
Source: National program announcements; estimates
The Threat
Why governments really care
A large enough quantum computer would break the encryption that protects almost everything: banking, secrets, the internet itself.
That machine doesn't exist yet. The danger is that adversaries are collecting encrypted data to decrypt later.
How quantum breaks today's encryption
- Today's locks: RSA encryption relies on factoring huge numbers being impossibly slow.
- Shor's algorithm: a quantum computer could factor those numbers efficiently.
- The lock opens: encrypted data, past and present, becomes readable.
- Harvest now: adversaries store encrypted data today to crack later.
Under a million qubits
Still needed to break RSA-2048 encryption on current estimates: fewer than one million noisy qubits running about five days (Google researcher Craig Gidney, May 2025), down from roughly 20 million in 2019. Today's machines have ~1,000 noisy qubits and cannot break real encryption: the largest key cracked so far was 22 bits, versus 2,048 in practice. The number required keeps falling as methods improve.
When does the encryption clock run out?
- no machine can break RSA (safe today)
- data stored to crack later
- RSA deprecated (NIST)
- vulnerable algos gone
- RSA broken in practice
Source: NIST timelines; industry estimates.
The world is already migrating
| Step | Status |
|---|---|
| NIST standards | in 2024, NIST finalised the first post-quantum encryption standards. |
| Deprecation | RSA-2048 is slated for phase-out by 2030-2035. |
| Migration | governments and banks are beginning the multi-year switch. |
| Crypto-agility | the goal: systems that can swap algorithms quickly. |
Reality Check
Hype versus reality
What quantum will and won't do
The hype versus the reality
| The hype | The reality |
|---|---|
| 'Solves everything' | it doesn't, only narrow problems |
| 'Here now' | useful machines are years away |
| 'Replaces classical' | never, for everyday tasks |
| 'AI supercharged' | largely unproven |
- Reality, narrow but real: chemistry, materials, optimisation.
- Reality, a 2030s payoff: for genuine advantage.
- Reality, a co-processor: works alongside classical machines.
- Reality, a present threat: to encryption, via 'harvest now'.
Three things quantum computing is not
| Myth | Reality |
|---|---|
| Not 'try all answers at once' | it explores superpositions, but you can't read them all; interference does the real work. |
| Not faster internet | 'quantum' networking is about unbreakable security, not speed. |
| Not conscious or magic | it is linear algebra on extraordinarily fragile hardware, nothing more. |
The realistic early use cases
- Chemistry and drugs: simulating molecules classical computers can't, for new materials and medicines.
- Materials science: batteries, catalysts, superconductors designed atom by atom.
- Optimisation: logistics, finance and scheduling, though the edge is still unproven.
- Cryptography: both the threat, and the quantum-safe defences it forces.
The right mental model: quantum is a specialised co-processor for problems classical computers find impossible, not a replacement for anything you use today.
Like a GPU for physics, it will sit alongside classical machines, not supplant them.
What to Watch
Signposts, not predictions
Four dials worth watching
| Dial | Measure | What it tells you |
|---|---|---|
| Logical qubits | Quality | the number that actually matters |
| Error rates | Fidelity | falling = real progress |
| PQC migration | Security | how fast the world switches |
| First advantage | Use case | a real problem, solved for real |
Source: as cited in this deck.
Five signposts for the next 24 months
- Do logical qubits scale? The real measure of progress, not physical count.
- Does anyone reach fault tolerance? The threshold to a useful machine.
- Does a real use case land? One problem solved better than any classical machine.
- How fast does PQC migration go? The security clock is already ticking.
- Does the funding hold? Quantum needs patient capital through a long winter.
Reference
Glossary and sources
Plain-language glossary
| Term | Meaning |
|---|---|
| Qubit | a quantum bit, can be 0 and 1 at once. |
| Superposition | being in multiple states simultaneously. |
| Entanglement | qubits linked so one's state depends on another's. |
| Logical qubit | one reliable qubit built from many physical ones. |
| Fault tolerance | computing correctly despite hardware errors. |
| Decoherence | the loss of quantum state to the environment. |
| Shor's algorithm | the quantum method that threatens RSA encryption. |
| Post-quantum crypto | encryption designed to resist quantum attacks. |
Quantum computing is over-hyped for this decade and under-appreciated for the next. The machine that changes the world doesn't exist yet, but the physics says it can, and that is enough to change everything now.
The five conclusions to take away
- A different tool, not a faster one: narrow power, not universal speed.
- Errors are the whole game: logical qubits matter; raw counts don't.
- Useful advantage is a 2030s story: patience is required.
- The security threat is present: 'harvest now, decrypt later' is real.
- First value in simulating nature: chemistry and materials, before anything else.
Sources and further reading
- Google Quantum AI: 'Meet Willow, our state-of-the-art quantum chip' (2024).
- IBM Quantum: fault-tolerant roadmap and processor disclosures (2024-25).
- Microsoft / Atom Computing: logical-qubit demonstrations (2024).
- NIST: Post-Quantum Cryptography Standards, FIPS 203/204/205 (Aug 2024).
- MarketsandMarkets: Quantum Computing Market forecast to 2030.
- National programs: US National Quantum Initiative, EU Quantum Flagship, others.
- The Quantum Insider: company roadmaps and milestone tracking (2025).
- McKinsey: Quantum Technology Monitor 2026 (Apr 2026); The Quantum Insider (Jun 2026); Quantum Computing Report news archive (2026); Gidney, RSA-2048 resource estimate (May 2025).
- Figures marked 'estimate' or 'illustrative' are compilations; the field moves fast.
Sources cited
MarketsandMarkets · Google · IBM · Company disclosures · IBM roadmap · The Quantum Insider · PitchBook · National program announcements · Company roadmaps · NIST · Microsoft / Atom Computing · McKinsey, Apr 2026 · Quantum Computing Report, 2026 · The Quantum Insider, Jun 2026 · Craig Gidney (Google), May 2025 · postquantum.com, Jun 2026 · IBM Quantum · NIST CSRC, Oct 2026
Keep reading
Research, not investment advice. Figures as of , from the sources cited.
Legal information and disclosures
These are my own views as of the update shown on this page. They can change without notice, and I have no duty to update them. Past performance does not indicate future results, and any investment can lose money.
This is for education only. It is not investment, legal, tax, medical or accounting advice, and it is not an offer to sell or a solicitation to buy any security in any jurisdiction. It does not consider your goals or circumstances.
Figures from company filings, public agencies and other third parties are believed to be reliable, but I have not independently verified them and cannot guarantee them. Forecasts and anything marked as an estimate rest on assumptions that may prove wrong.
Please do not copy or republish this material without my written consent.