Quantum computing is experiencing a breakthrough moment as Google’s Willow chip achieves scalable quantum error correction, sparking investor frenzy and government investment worldwide. Explore the future of fault-tolerant quantum computers, key technologies (superconducting, trapped ion, neutral atom, photonic, topological), commercialization challenges, and why this may be the next big revolution in computing and AI.
On December 9, 2024, Google published a paper in Nature that made quantum physicists lose their minds in a very controlled, peer-reviewed way.
Their Willow chip—105 qubits of superconducting magic cooled to nearly absolute zero—did something that sounds boring until you understand it: they made quantum error correction work exponentially better as they scaled up.
For three decades, quantum computing had a really embarrassing problem. It's like if every time you tried to make your startup bigger, it got worse at the thing it was supposed to do. You hire more engineers and ship fewer features. You raise more money and burn faster. You add more qubits and... you get more errors.
Not anymore.
Willow demonstrated 2.14× error reduction every time they increased the "code distance" (basically, how much error correction they threw at the problem). The distance-7 code hit 0.143% logical error per cycle while coherence times improved 5× over their previous chip. This is below the threshold. The thing works. Scaling works. Physics works the way the textbooks said it should.
And everybody started acting like it's 2007 and Steve Jobs just pulled the iPhone out of his pocket.
Here's the thing about quantum computers that nobody tells you upfront: they've been simultaneously five years away and revolutionary for about 30 years now.
They're Schrödinger's investment thesis—both the future of computing AND a science experiment, existing in superposition until someone checks the quarterly earnings.
But Google's Willow breakthrough matters because it validates the core bet: you can build a fault-tolerant quantum computer by scaling up, not just optimizing what you have. This is the difference between "we need to make our qubits 1000× better" (hard) and "we need to make 1000× more qubits" (also hard, but a different kind of hard that involves money and engineering rather than new physics).
The market certainly thinks something changed. Let me show you some absolutely bonkers numbers:
These are AMC-GameStop-2021-meme-stock numbers. Analysts are using phrases like "absolutely scream bubble." Which... yeah. But also, bubbles can be right about the future and wrong about the timing. The dot-com bubble was correct that the internet would be huge. It was just early on most of the companies.
So let's figure out: Is quantum computing the iPhone or the Segway? Are we in 2007 or 2000?
Here's where quantum gets interesting from an investment perspective: nobody knows which technology will win, so everybody's hedging their bets.
There are five main ways to make a qubit (a quantum bit, the basic unit of quantum computing), and each one is basically a different bet on physics:
The vibe: This is the incumbent. The safe play. The "nobody ever got fired for buying IBM" of quantum.
How it works: You make tiny circuits out of superconducting materials, cool them to 10-20 millikelvin (that's 0.01-0.02 degrees above absolute zero, colder than outer space), and use microwave pulses to manipulate quantum states.
The good: Most mature tech. IBM has 22+ quantum computers running. Google just proved error correction works. You can already rent time on these things through AWS Braket.
The bad: Each qubit costs ~$10,000 in supporting infrastructure. Coherence times are only 30-100 microseconds (that's 0.00003-0.0001 seconds), meaning you have to do your quantum computation really, really fast before everything falls apart. Oh, and you need a dilution refrigerator the size of a chandelier.
Current leader: IBM (1,121-qubit Condor), Google (105-qubit Willow with the error correction breakthrough)
Market cap proxy: IBM is too big to be a pure play, but Rigetti is public at ~$500M market cap (though it changes daily by like 40%).
The vibe: The perfectionist. High quality, high fidelity, but moves slower than the superconducting crew.
How it works: You trap individual atoms (ytterbium or calcium ions) using electromagnetic fields, then use lasers to manipulate them. The atoms themselves ARE the qubits.
The good: 99.9% two-qubit gate fidelity (vs. 99.5% for superconducting). Coherence times in milliseconds instead of microseconds—that's 100-1000× better. All-to-all connectivity means any qubit can talk to any other qubit directly, which is huge for reducing overhead.
The bad: Gate operations are 10,000× slower than superconducting qubits. Scaling beyond a few hundred qubits is hard because you need to control each atom with individual lasers, and laser control systems are complicated.
Current leaders: IonQ ($4B market cap, $43M revenue, big Air Force contract), Quantinuum (private, just raised at $10B valuation with JPMorgan and NVIDIA backing)
My take: IonQ is basically running the DoorDash playbook—go public via SPAC, use the capital to consolidate the market, vertically integrate everything. They've spent $1+ billion on acquisitions in the past year. That's... aggressive for a company that did $43M in revenue. But if they're right about the technology, it's genius.
The vibe: The dark horse nobody was paying attention to until suddenly they had more qubits than everybody else.
How it works: Similar to trapped ions, but you use neutral atoms (rubidium or cesium) held in place by optical tweezers. Think of it as using light to create an egg carton, then putting an atom in each cup.
The good: They scaled to 1,180 qubits in 2023—first system to break 1,000. Coherence times of 40 seconds (that's 400,000× better than superconducting). Here's the kicker: because you're using actual atoms from the periodic table, every qubit is perfectly identical by the laws of physics. You don't have manufacturing defects.
The bad: Gate fidelity is slightly lower at 99.5-99.6%. Still needs to prove out at scale.
Current leaders: Atom Computing (private, partnered with Microsoft, demonstrated 24 entangled logical qubits in November 2024—world record), QuEra (private, available on Amazon Braket)
Why this is interesting: Microsoft partnered with Atom Computing, which is significant because Microsoft is the only hyperscaler that doesn't have its own internal quantum hardware program. They're basically saying "we'll build the software stack, you build the hardware." Classic Microsoft play—they did the same thing with Intel in the PC era.
The vibe: The contrarian all-in bet. Skip the science experiment, go straight to utility scale or die trying.
How it works: Use photons (particles of light) as qubits, process them through silicon photonic chips manufactured at semiconductor fabs.
The good: Leverages existing telecom and semiconductor manufacturing infrastructure. Room temperature operation (no expensive cryogenics). Photons don't interact with electromagnetic interference. Can manufacture on 300mm wafers at GlobalFoundries.
The bad: Photon loss is a massive problem. Detection efficiency is hard. Nobody has proven this works at scale yet.
Current leader: PsiQuantum (private, $3.15B raised, most recent valuation unknown but likely $5B+)
The absolutely wild thing: PsiQuantum raised $620 million from the Australian government—the single largest quantum investment globally—to build a million-qubit fault-tolerant quantum computer by 2027-2029. They're not building a 100-qubit system first. They're not doing NISQ (Noisy Intermediate-Scale Quantum). They're going straight for the kill shot: a utility-scale machine with a million qubits that can run actual useful algorithms.
This is either visionary or insane. Probably both.
The vibe: Highest risk, highest reward. The "we're going to do this the right way even if it takes 15 years" play.
How it works: Use exotic quantum properties of matter (Majorana Zero Modes in topoconductor materials) to create qubits that are protected from errors at the hardware level.
The good: If it works, you could potentially fit 1 million qubits on a single chip without massive error correction overhead. The errors are protected by topology—the shape of the quantum state itself—rather than by constantly measuring and correcting.
The bad: It's really, really hard. Microsoft has been working on this for over a decade. They just announced their first demonstration in February 2025. No commercial systems exist. This is years away.
My take: Microsoft is playing 4D chess while everyone else plays regular chess. If topological qubits work, Microsoft leapfrogs everybody. If they don't, Microsoft has partnerships with all the other approaches anyway (Azure Quantum supports superconducting, trapped ion, neutral atom, etc.). They literally cannot lose as long as someone makes quantum work.
Here's where I'm supposed to tell you which quantum computing approach will dominate the 2030s.
I have no idea. Nobody does.
NVIDIA's NVentures—which is basically NVIDIA's "we're smart money" signal—invested in QuEra (neutral atom), Quantinuum (trapped ion), AND PsiQuantum (photonic). They're hedging across three different modalities because even NVIDIA, which employs some of the smartest chip people on Earth, doesn't know which physics will win.
But here's what we DO know:
Superconducting is the safe bet for 2025-2028. It's the most mature, has the most companies working on it, and has the commercial infrastructure. If you need a quantum computer in 2026 for research or pilot programs, you're probably using superconducting.
Trapped ion is the quality bet. Higher fidelity, better coherence, and current commercial traction. IonQ and Quantinuum are the two best-capitalized pure quantum plays, and both are trapped ion. That's not a coincidence.
Neutral atom is the scaling bet. If we need 100,000+ physical qubits to make useful quantum computers (we probably do), neutral atoms have the clearest path to get there because of inherent qubit uniformity and long coherence times.
Photonic is the "skip levels" bet. If PsiQuantum pulls off their million-qubit system using semiconductor manufacturing, they could leapfrog everyone. But it's the highest-risk approach.
Topological is the "we win by 2035" bet. If Microsoft's approach works, they win the long game. But it's too early to know if it works at all.
Let me walk you through the math on why quantum computing valuations are simultaneously justified and insane.
McKinsey says quantum computing is a $28-72 billion market by 2035, growing to $198 billion by 2040 when you include quantum sensing and communications.
BCG says it's $450-850 billion in value creation by 2040, with $90-170 billion in revenue to quantum providers.
The use cases are genuinely transformative:
If even ONE of these applications hits in a meaningful way by 2030-2032, the total addressable market is measured in hundreds of billions.
Current quantum computing revenue is ~$1 billion industry-wide. The three public quantum companies combined did maybe $55 million in 2024 revenue.
That's a 1,000× gap between current revenue and 10-year projections.
Zero drugs have been developed using quantum computers. Zero production financial models are running on quantum. Zero supply chains have been optimized by quantum. The one commercial deployment with genuine quantum advantage? Random number generation for JPMorgan's cryptography. That's it. After five years of investment.
The commercialization gap is enormous.
Current quantum computers can't simulate molecules larger than 10-40 atoms with acceptable accuracy. Drugs are 30-50+ atoms plus complex protein interactions. You need fault-tolerant quantum computers with 1,000+ logical qubits to do anything useful.
And here's the overhead problem: current implementations require 1,000-10,000 physical qubits per logical qubit. So to get 1,000 logical qubits, you need 1-10 million physical qubits.
At $10,000 per superconducting qubit in infrastructure costs, that's $10 billion to $100 billion in hardware. For ONE machine.
Does that math work? Eventually, yes—if quantum computers can solve multi-hundred-billion-dollar problems, spending $10B on a machine makes sense. But the timeline is uncertain, and capital requirements are massive.
The market is pricing in the bull case happening on an aggressive timeline (2029-2032) while mostly ignoring the risks:
Analysts are putting probabilities at:
So there's basically a coin flip on whether this works in the next 5-7 years.
Here's the part that makes quantum different from your typical VC-backed deep tech:
Governments are going ALL IN.
This isn't venture capital looking for 10× returns. This is nation-state competition. Quantum computing intersects:
The game theory is fascinating. If China gets fault-tolerant quantum computers first and you don't, they can potentially break your encrypted communications, accelerate their drug discovery, optimize their logistics networks, and develop better AI. That's an unacceptable strategic position.
So governments are funding quantum computing as a national security imperative, not a commercial ROI calculation. This means the technology gets developed regardless of near-term business models.
For investors, this is actually bullish. Government contracts provide:
IonQ's $54.5 million Air Force Research Lab contract is a perfect example. That's more than their entire 2024 revenue from commercial sources. It stabilizes their burn rate and gives them breathing room to actually build the technology.
If you're a startup thinking about quantum, here's what's interesting: you probably shouldn't build a full quantum computer.
The SBIR/STTR landscape reveals what actually gets funded:
Components and enabling technologies, not full systems.
NSF Phase I awards are up to $305K for 6-12 months. Phase II goes up to $1.25 million for 24 months. DOE SBIR Phase II is even bigger at $1.6 million. These are meaningful amounts of non-dilutive capital.
But look at what wins:
See the pattern? Nobody's getting SBIR funding to build a competing quantum computer from scratch. That's too ambitious, requires too much capital, and has too many well-funded incumbents.
The opportunities are in:
The picks-and-shovels strategy works because you can sell to multiple quantum platforms regardless of which modality wins. Plus, acquisition risk is lower—full-stack quantum companies like IonQ are actively acquiring these capabilities (they just spent $1+ billion on acquisitions).
If you're writing an SBIR proposal, evaluation criteria emphasize:
The mistake most applicants make: proposing something too ambitious. "We're going to build a 1000-qubit quantum computer" loses to "We're going to reduce two-qubit gate error rates by 10× using this novel control system architecture."
If you're a VC or angel thinking about quantum, here's my framework:
This is not a 2027 exit. Even the aggressive roadmaps (IBM's 200 logical qubits by 2029, IonQ's 1,600 logical qubits by 2028) are just getting to fault tolerance. Commercial applications at scale are 2030s.
Can you wait that long? Do your LPs have that patience? If not, don't invest in quantum.
NVIDIA's NVentures strategy is right—invest across 2-3 modalities. Unless you're a quantum physicist who has strong conviction about which approach wins, spreading across superconducting, trapped ion, and neutral atom (or photonic as a wild card) makes sense.
75%+ of Q1 2025 funding went to full-stack companies. The thesis: the software stack will be built around the winning hardware, and full-stack players can capture the most value.
But 25% going to components isn't nothing. Those are acquisition targets or businesses that can sell to multiple platforms.
Prioritize companies with meaningful government revenue. IonQ's $54.5M Air Force contract, PsiQuantum's $620M from Australia, and quantum hub funding are massive de-risking signals.
Plus, government contracts often convert to commercial relationships. The Department of Defense doesn't just want to research quantum computing—they want to deploy it for logistics, cryptography, and simulation. Those are real use cases with real budgets.
Normal SaaS metrics don't work for quantum. Revenue is tiny, losses are massive, and nobody's profitable.
Instead, track:
When companies start showing multiple Fortune 500 pilot programs converting to production deployments, THAT'S the signal.
Quantum talent clusters matter:
If you're starting a quantum company, these clusters offer talent pools, government support, and ecosystem effects.
IPO? Possible, but difficult without profitability path. IonQ succeeded, Rigetti struggled initially, D-Wave has been volatile.
SPAC? Market cooled significantly after 2021-2022.
Acquisition? Most likely outcome for components/software. Full-stack plays might IPO if they hit milestones. Big tech (Google, Microsoft, Amazon, IBM) and defense contractors (Lockheed, Northrop) are potential acquirers.
Acquihire? Common for talent, less common for technology given sector immaturity.
I've now spent three weeks researching quantum computing and talking to quantum physicists, VCs, and founders. Here's my take:
This is real. But it's early.
Google's error correction breakthrough is legitimate scientific progress. We're past the point where someone can say "quantum computing might not physically work." It works. Scaling works. The physics validates.
But we're probably in 2001, not 2007.
The roadmaps converging on 2029-2030 for fault-tolerant systems with hundreds of logical qubits might be aggressive. Could slip to 2032-2035. And even when fault-tolerant systems exist, enterprise adoption takes time. Figure another 2-5 years for real commercial deployments at scale.
So meaningful revenue probably comes in the 2032-2037 window. Call it 7-12 years from now.
The valuations are insane by traditional metrics but not crazy if you squint.
IonQ at 90× forward P/S is objectively bonkers. But if you believe they're building a $10B+ revenue business by 2035 (reasonable if quantum works), then today's $4B market cap is actually cheap. You're paying 0.4× forward revenue on a 10-year view.
The bet is pure binary: either quantum computing becomes massive and these companies 10× from here, or quantum winter hits and they go to zero. There's not much middle ground.
The meta-strategy is diversification.
Can't pick the winning qubit modality? Buy the index: invest in superconducting, trapped ion, AND neutral atom.
Can't pick the winning company? Buy software and components that sell to everyone.
Not convinced on timing? Allocate small position sizes and be prepared to add on technical milestones.
For founders: build picks and shovels.
Unless you have $500M+ in funding and a decade-long timeline, don't build a full-stack quantum computer. You're competing with Google, IBM, Amazon, Microsoft, and heavily-funded startups with government backing.
Build the control systems. The error correction software. The cryogenic cooling solutions. The quantum networking protocols. The post-quantum cryptography. These are venture-scale businesses that can exit in 5-7 years to the full-stack players.
For LPs and allocators: small allocation, long patience.
If you have a deep tech bucket, quantum deserves 5-10% of it. Not more unless you have strong conviction and specialized expertise.
Expect 10-15 year hold periods. Expect volatility. Expect 60-70% of quantum investments to go to zero. But the winners could be 50-100× if this plays out bull case.
This is option value on computing's next paradigm. You're paying a premium for exposure to potentially transformational technology. Sometimes that's worth it.
Quantum computing is simultaneously overhyped and underhyped.
Overhyped: The stock price movements, the "quantum supremacy" headlines, the promise that quantum computers will cure cancer by 2027.
Underhyped: The genuine scientific progress, the government strategic commitment, the fundamental physics validating that this will eventually work, and the magnitude of impact when it does.
Google's Willow chip proved that scaling quantum computers isn't just possible—it works exponentially better as you scale. That's a big deal. Maybe the biggest deal in computing since the transistor.
But scaling from 105 qubits to 100,000+ qubits, from research demonstrations to commercial deployments, from pilot programs to production systems—that's going to take time. Longer than the bulls think. Maybe less time than the bears think.
The smart money is diversified, patient, and focused on technical milestones rather than revenue multiples.
The dumb money is chasing 4,000% gains on quantum stocks with no revenue model.
The really smart money? They're building the components and enabling technologies that every quantum computer needs, getting SBIR grants to de-risk development, and positioning to be acquired by the full-stack winners in 3-5 years.
We're living through the birth of a new computing paradigm. It's messy, uncertain, expensive, and absolutely fascinating.
Just don't bet the whole portfolio on it working by 2028.
