Most of this book has been about the durable physics and engineering of quantum computing: what a qubit is, how superposition and entanglement are put to work, how errors creep in, and how an instrument rack actually controls a chip held near absolute zero. Those fundamentals change slowly. The field around them does not. Quantum computing is one of the fastest-moving areas in all of technology, and the picture shifts month to month as new processors, new error-correction results, and new advantage claims arrive.
This chapter is the snapshot. It steps back from the physics to ask where the field stands today, what the open questions are, what the industry is actually saying, and which milestones over the period from 2021 to 2026 mark real progress rather than press-release noise. Because this is the chapter most exposed to change, every dated claim carries a numbered reference flagged for verification before publication. By the time you read this, some of these records will have been broken. That is the point: treat the specifics as a starting line, not a finish line.
The hard questions in quantum computing are no longer "does any of this work at all." Small machines exist, they run real circuits, and they produce results that classical simulation struggles to reproduce. The questions now are about scale, and they are stubbornly practical. SeeQC's John Levy framed the challenge as a set of engineering problems that a serious company actually cares about, and that framing still holds. The open questions cluster around a few themes.
Can you build a quantum computer scaled to the complexity of a problem a large enterprise cares about? A machine that factors small numbers or simulates a four-atom molecule is a demonstration. A machine that simulates a catalyst, prices a derivative portfolio, or breaks a real cryptographic key is a product. The gap between those two is enormous, and it is measured in qubits, fidelity, and runtime all at once.
Can you scale it from an energy perspective? A modern superconducting system spends most of its power not on the qubits but on the dilution refrigerator and the room-temperature electronics that drive it. As qubit counts climb into the thousands and beyond, the heat budget becomes a first-order constraint, not an afterthought.
Can you scale it from a cost and system-complexity perspective? Today a quantum computer is a room full of racks, cables, and cryogenics that bridge low temperature to high temperature and back, analog to digital and digital to analog, over and over. Each of those conversions is a place for noise to enter and cost to accumulate.
Can you scale the control technology itself? Microwave control of superconducting qubits is both analog and energetic. Every control line injects heat into a system whose entire purpose is to stay cold. Cable count is the blunt version of the same problem. A superconducting qubit can require several coaxial lines for control and readout. Multiply that by a million qubits and the cabling alone becomes physically impossible with today's approach. This is why "all-digital" control, qubit multiplexing, and cryogenic control electronics are such active areas of work.
These questions describe the NISQ era, short for Noisy Intermediate-Scale Quantum, a term coined by Caltech's John Preskill. NISQ machines have tens to low thousands of physical qubits and no full error correction, so noise limits how deep a useful circuit can run. They are real, and they are useful for research, benchmarking, and early algorithm development. They are not yet solving climate change, turbocharging financial forecasting, or designing drugs end to end. Those commercial applications are further down the road, on the far side of fault tolerance. The framing that matters is this: the field has answered "can it work," and it is now grinding through "can it scale," which is a much harder and much more expensive question.
We have been here before, in a sense. Classical computing made the same march from the transistor to the integrated circuit to the microprocessor before anyone could buy a useful computer. Quantum computing is somewhere early in that arc, taking racks of laboratory equipment and slowly turning them into something an engineer can deploy.
The first edition of this book closed with a scrapbook of news clippings, dozens of headlines collected to make a single point: quantum computing was further along, and moving faster, than most readers assumed. That point is even truer now, and the tone of the coverage has matured. The early years were dominated by wonder and hype. The current industry voice is more measured, more focused on engineering, and noticeably more concerned with timelines and risk.
Three threads run through almost everything serious that is written about the field today.
The first thread is the shift from physical qubits to logical qubits as the metric that matters. For years the headline number was the raw qubit count, and vendors raced to announce the biggest chip. That race has not stopped, but the conversation has moved on. A thousand noisy physical qubits cannot do what a handful of error-corrected logical qubits can, and the industry knows it. Coverage now leads with logical-qubit demonstrations and error-correction results, because those are the numbers that predict whether a machine will ever be useful.
The second thread is the security clock. Government cyber agencies on both sides of the Atlantic have moved from speculation to scheduling. The message is consistent: a cryptographically relevant quantum computer does not exist yet, but data stolen and stored today could be decrypted once one does, so the migration to quantum-resistant cryptography has to start now. This "harvest now, decrypt later" risk has turned post-quantum cryptography from an academic topic into a board-level concern, and it is the single most common reason a non-physicist executive now pays attention to quantum at all.
The third thread is the move from laboratory to industry. The clearest signal is who is spending money. Large enterprises in finance, pharmaceuticals, materials, logistics, and aerospace have signed multi-year partnerships with quantum hardware vendors, and several have installed on-premises machines or reserved dedicated cloud access. Health systems, banks, and chemical companies are building internal teams whose job is to be ready when the hardware arrives. The applications are still mostly exploratory, but the budgets are real, and that is a different kind of evidence than a press release.
A useful way to read any quantum headline is to ask which of three things it actually reports. Is it a hardware milestone (more qubits, better fidelity, a new error-correction result)? Is it a commercial milestone (a partnership, an installation, a funding round)? Or is it an application claim (this machine solved this problem)? The first two are usually verifiable and meaningful. The third deserves the most scrutiny, because "quantum advantage" claims are routinely narrowed or overturned when classical algorithms improve. Healthy skepticism is not cynicism here. It is how the field polices itself.
The period since the first edition has been the most productive stretch in the field's history. The milestones below are grouped by theme rather than strict chronology, because the threads advanced in parallel. Dates and figures are drawn from the cited sources and should be verified against the original announcements before publication, since this area changes quickly.
Error correction crossed the threshold. The single most important result of the period came in December 2024, when Google Quantum AI reported that its Willow superconducting processor had operated a surface code below the error-correction threshold for the first time. [1] In plain terms, they showed that making the logical qubit bigger made it better: each step up in code distance suppressed the logical error rate by more than a factor of two, with a distance-7 code spread across 101 physical qubits achieving a logical lifetime longer than its best individual physical qubit. [1] Below-threshold operation is the property fault tolerance requires, and before Willow no processor had clearly demonstrated it in a surface code. It moved a textbook promise into a measured laboratory fact.
Logical-qubit counts climbed. Demonstrating one good logical qubit is a milestone. Demonstrating many, and entangling them, is the next. In late 2024 Microsoft, working with Atom Computing on neutral-atom hardware, reported creating and entangling 24 logical qubits, encoded across roughly 112 physical atoms, the largest number of entangled logical qubits reported to that point. [2] Quantinuum, on trapped-ion hardware, demonstrated 12 logical qubits with high fidelity in the same era. [2] The trend is unmistakable: the frontier metric has shifted from how many physical qubits a chip has to how many usable logical qubits a system can run.
Processors got larger. The raw-qubit race did not stop, even as attention moved to logical qubits. Both IBM and Atom Computing crossed the symbolic 1,000-physical-qubit line. IBM's Condor reached 1,121 superconducting qubits, the first superconducting processor past that mark. [3] Atom Computing fielded a neutral-atom system with a 1,225-site array populated by roughly 1,180 atoms, and Pasqal trapped more than 1,100 atoms in its neutral-atom processor. [3] These large machines are less about running useful algorithms today and more about proving that the manufacturing and control challenges of scale can be met.
Quantum-advantage claims advanced, and were contested. Random circuit sampling remained the main proving ground for raw quantum advantage. Google reported in October 2024 that its 67-qubit Sycamore processor performed a sampling task it estimated would take the Frontier supercomputer thousands of years. [4] China's Zuchongzhi 3.0 reported comparable or stronger claims against the same classical baseline. [4] At the same time, classical-simulation methods kept improving, repeatedly chipping away at earlier advantage claims, which is exactly why advantage is best understood as a moving contest rather than a one-time victory. [4]
Roadmaps got specific. A notable change in tone is that the leading vendors now publish dated, engineering-grade roadmaps toward fault tolerance rather than vague aspirations. IBM laid out a path to a fault-tolerant machine it calls Starling, targeted for 2029 with on the order of 200 logical qubits running 100 million gates, supported by interim processors and new error-correcting-code work along the way. [5] Quantinuum, IonQ, and others published their own dated targets for fault-tolerant or cryptographically relevant machines later in the decade. [5] These timelines are commitments, not guarantees, but their specificity is itself a sign of a maturing field.
A new hardware bet went public. In February 2025 Microsoft announced Majorana 1, a processor built on topological qubits, an approach designed to make error protection partly a property of the hardware itself rather than something added entirely in software. [6] The topological approach remains scientifically contested and unproven at scale, but its public debut widened the set of serious hardware bets beyond superconducting circuits, trapped ions, and neutral atoms.
Cryptography got its standards. The most consequential non-hardware milestone touches every connected device on earth. In August 2024 the U.S. National Institute of Standards and Technology finalized its first post-quantum cryptography standards: FIPS 203 (ML-KEM) for key exchange, FIPS 204 (ML-DSA) for digital signatures, and FIPS 205 (SLH-DSA) as a hash-based signature fallback. [7] NIST added a fifth algorithm, HQC, in 2025 as a backup based on different mathematics. [7] These standards exist because the threat is taken seriously enough to act on before the machine that justifies them is built. National-security guidance now sets firm migration deadlines around 2030. [7]
The throughline across all of these milestones is consistency of direction. Error rates fell, logical-qubit counts rose, processors grew, roadmaps sharpened, and the defensive infrastructure of post-quantum cryptography moved from draft to standard. None of it amounts to a useful, fault-tolerant, commercially decisive quantum computer yet. All of it points the same way.
Going Deeper - Reading a quantum headline critically
When a new "world record" lands, three questions separate signal from noise. First, is it peer-reviewed or a press release? A Nature paper and a keynote slide are not the same evidence. Second, what exactly was measured, and against what baseline? An advantage claim is only as strong as the classical algorithm it beat, and those algorithms improve. Third, is the result physical, logical, or commercial? More physical qubits is the weakest of the three signals, a new below-threshold logical result is among the strongest, and an application claim sits in between and usually needs the most scrutiny. Apply those three filters and most of the hype falls away, leaving the genuine progress underneath.
BNC in Practice - The instrument bench behind the headlines
Every milestone in this chapter rests on a foundation of precise classical instrumentation. A below-threshold error-correction result depends on low-noise control signals, tightly synchronized timing across many channels, and clean readout. Larger processors multiply the demand for stable references, precise pulse generation, and accurate signal sources feeding the control electronics. Berkeley Nucleonics builds signal generators, timing and synchronization instruments, and reference sources used in research and test environments of this kind. The quantum processor gets the headline. The measurement and control rack around it is what makes the headline reproducible. Match any specific instrument to the frequency range, timing resolution, and channel count your experiment needs, and verify the specifics against the current datasheet before you commit.
Take it interactively. The quiz lives on its own page with hidden answers - write your attempt first (even four characters works), then reveal. Self-graded. About 10 minutes.
Or read the questions and answers inline below (preserved for print and offline use).
[1] Google Quantum AI, "Quantum error correction below the surface code threshold," Nature, December 2024; distance-7 surface code on 101 physical qubits, logical error suppression factor exceeding 2 per code-distance step. Verify before publication.
[2] Microsoft and Atom Computing demonstration of 24 entangled logical qubits on neutral-atom hardware (roughly 112 physical atoms), late 2024; Quantinuum 12-logical-qubit demonstration on trapped-ion hardware. Verify before publication.
[3] IBM Condor superconducting processor (1,121 qubits); Atom Computing neutral-atom system (1,225-site array, ~1,180 atoms); Pasqal neutral-atom processor (>1,100 atoms). Verify before publication.
[4] Google 67-qubit Sycamore random circuit sampling result (October 2024); China's Zuchongzhi 3.0 random circuit sampling claims; ongoing improvements in classical simulation that contest advantage claims. Verify before publication.
[5] IBM roadmap to the fault-tolerant "Starling" system (target 2029, ~200 logical qubits, 100 million gates); dated fault-tolerance and cryptographic-relevance roadmaps from Quantinuum, IonQ, and others. Verify before publication.
[6] Microsoft "Majorana 1" topological-qubit processor announcement, February 2025. Topological approach remains unproven at scale. Verify before publication.
[7] NIST finalized post-quantum cryptography standards, August 2024: FIPS 203 (ML-KEM), FIPS 204 (ML-DSA), FIPS 205 (SLH-DSA); HQC selected as a fifth algorithm in 2025; national-security migration deadlines around 2030. Verify before publication.