Quantum problem solved on ordinary PC: breakthrough by Flatiron Institute

American Scientists Solved an 'Impossible' Quantum Problem Using an Ordinary PC

Researchers at the Flatiron Institute (USA) have disproven the notion that only a quantum computer is needed to simulate quantum systems with complex 'spin glasses'. Using a classical computer with specialized compression algorithms, they achieved results comparable to the D-Wave Advantage2, proving that classical systems have not yet exhausted their potential.

---

Classical PC vs. Quantum Monster: How Flatiron Institute Shattered D-Wave's $7.5 Billion Argument

Analytical review as of May 30, 2026

[The Gist]: What's Really Happening

On May 28, 2026, a group of physicists from the Center for Computational Quantum Physics at the Flatiron Institute (New York) published results in the journal Science that called into question the foundation supporting D-Wave Quantum Inc.'s (NYSE: QBTS) $7.53 billion market capitalization.

Joseph Tindall, Miles Stoudenmire, and their colleagues took on the problem of simulating quantum spin glasses—chaotically frozen magnetic moments of atoms—which in March 2025 was presented by D-Wave as 'quantum supremacy', and solved it on an ordinary computer using data compression algorithms.

The key figure that didn't make mainstream headlines: Tindall performed part of the calculations on an ordinary laptop. Yes, the very device that costs $1,000–2,000 at Best Buy. For the most complex 3D configurations, a workstation with a powerful GPU was required—but that's still a classical system, containing not a single qubit.

Insider insight: This is not a 'technical achievement'. It is an architectural blow to the entire philosophy of quantum supremacy. If classical algorithms with tensor networks and belief propagation can mimic quantum annealing with accuracy comparable to the D-Wave Advantage2, then the unique value proposition of quantum computers for optimization problems collapses. And it is precisely on these problems that D-Wave has built its entire business.

Timeline and Context

To understand the scale, we need to look at the sequence of events:

March 2025: D-Wave publishes a paper in Science claiming to have simulated the dynamics of spin glasses on a system with over 5,000 qubits on the Advantage2 processor. The company asserts that reproducing these results on the classical supercomputer Frontier would require nearly 1 million years of computation.

Late 2025: D-Wave begins aggressive promotion into the defense market. In January 2026, a partnership is announced with Anduril Industries (Palmer Luckey's company, valued at $14 billion) and Davidson Technologies. Results: 10x speedup for US missile defense solutions, 9–12% improvement in interception, enabling an additional 45–60 missiles to be shot down out of 500.

QBTS stock soars 270% over the year, peaking around $46 in its 52-week range.

May 27, 2026: Tindall's group publishes their version in Science. The classical modeling results are no worse, and in some tests even better, than those of the D-Wave Advantage2.

May 28, 2026: D-Wave issues an official rebuttal—a detailed technical counterargument—claiming that the classical algorithm fails on the most complex topologies (bicliques, 3D cubic lattices with strong coupling) and does not reproduce the full set of physical observables.

May 30, 2026: Here we are. The truth lies somewhere in between, but the market has yet to realize the implications.

Who Wins and Who Loses

Winners

• Simons Foundation (Flatiron Institute): Funded by the $10 billion foundation created by Renaissance Technologies co-founder Jim Simons, the Flatiron Institute has just proven that its investment in fundamental mathematics (the ITensor library) pays off at the highest level. Tindall used ITensor—an open-source library for tensor networks developed at the institute itself.

• Classical Computing Industry (NVIDIA, AMD, Intel): Flatiron's results show that software can be more important than hardware. NVIDIA (market cap $2.5 trillion) gains an additional argument: their GPUs on a basic workstation can compete with dedicated quantum systems. Tindall's algorithms were optimized for GPU acceleration.

• Research Labs Without Quantum Computer Budgets: Now universities and startups can simulate spin glasses and optimization problems without access to a $15 million D-Wave quantum system. The entry barrier drops from millions to thousands of dollars.

Losers

• D-Wave (QBTS): This is the main victim. Shares have already fallen 8% on the day of the news publication (from $22.13 to $20.35). But the main blow is to the narrative. D-Wave sells not so much hardware as the story of 'quantum supremacy in optimization.' If classical algorithms can do the same (albeit with some limitations), why should customers pay a premium? Contracts with the Pentagon through Anduril won't be canceled—the military needs real-time speed, and the 10x speedup remains a fact. But the commercial segment (logistics, finance, pharma) will now wonder: 'Maybe a good server with a GPU is enough for us?'

• Venture Capital Funds That Invested in Quantum Optimization Startups: If classical algorithms continue to catch up with quantum ones, exit strategies for 30+ startups in this niche (QC Ware, Zapata Computing, and others) become problematic. Their valuations were based on the assumption of a 'quantum gap' that is now narrowing.

• Academic Groups That Built Careers on 'Quantum Supremacy': Flatiron's results are a methodological challenge. Tindall and Stoudenmire directly stated: 'Every time we see such claims, we are a bit skeptical. Did you try this? And this?' This is a blow to the reputation of those who rushed to loud conclusions.

What the Media Aren't Telling You

Insight #1: D-Wave Knew About This 2 Months Before Publication—and Couldn't Do Anything

Tindall's paper was posted on arXiv (preprint 2503.05693) at the end of March 2025. D-Wave had 14 months to either refute the results or release new benchmarks that would be unattainable by classical methods.

Their official rebuttal from May 26, 2026 contains technically valid arguments—yes, BP-TNS fails on biclique graphs and some highly frustrated 3D spin glasses. But the company could not, in 14 months, demonstrate a problem that classical methods could not solve at all. This is a worrying signal. If you are a quantum company and cannot stay ahead of classical algorithms with a one-year lag, your technological barrier is low.

Insight #2: The Paper Doesn't 'Refute' Quantum Supremacy—It Redefines It

The media write 'scientists refuted quantum supremacy.' No. They showed that a specific problem—simulating spin glasses on specific topologies (square, cube, diamond)—does not require a quantum computer. D-Wave correctly points out that their own supremacy claims referred to a broader class of problems.

But here lies the main problem for D-Wave: their Advantage2 is built on a quantum annealing architecture that is originally designed for spin glasses and optimization problems. This is their 'killer feature.' If classical methods catch up with them on home turf, where is the real advantage?

Insiders know: D-Wave is preparing a full Advantage2 with 7,000 qubits (currently a prototype with 1,200+). But if at every step they are overtaken by Tindall's team with a new version of ITensor, the race turns into an endless cat-and-mouse game where the cat (classical computing) has an unlimited software budget.

Insight #3: No One Talks About the 'Free Lunch'—Belief Propagation Has Limitations

Belief propagation is an algorithm from the 1980s, adapted for quantum systems. It is very fast but approximate. Tindall himself admits: 'It is a slightly more approximate method than others.'

D-Wave's verdict that BP-TNS fails on strongly coupled 3D spin glasses and bicliques is technically correct. But Flatiron never claimed a universal solution. They showed that for a wide class of practically important problems (logistics, routing, portfolio optimization), classical methods may be sufficient. And for 'hard' problems—well, they may be so hard that even D-Wave cannot solve them with sufficient accuracy.

Forecast: Next 30 Days and 90 Days

Next 30 Days

• June 2026: D-Wave will be forced to publish new benchmarks on its full Advantage2 (1,200+ qubits) for problems where Flatiron's BP-TNS is guaranteed to break—likely strongly coupled biclique graphs. If D-Wave stays silent, investor confidence will drop.

• IEEE Quantum Week Conference (June): A direct discussion between Tindall and D-Wave representatives. This will be the main event. I bet Tindall will show an extension of his method to problems D-Wave considered 'safe.'

• Stock Reaction: QBTS may test the support level of $15–17 (another 20–25% drop from the current $20.35). Media noise around 'refuting quantum supremacy' will create short-term pressure.

Next 90 Days

• August 2026: We will see at least 2–3 major corporate clients of D-Wave (likely from the Fortune 500 in logistics) freeze or reconsider contracts for quantum services. If classical is 'good enough,' why pay $50–100k per month for access to Leap (D-Wave's cloud service)?

• ITensor 4.0 Release: The Flatiron team will release a new version of the library with optimizations for spin glass problems. This will make Tindall's results reproducible by any engineer with a good GPU. Open-source is a weapon against commercial quantum systems.

• Consolidation in the Quantum Optimization Sector: Startups without hardware protection (software-only or emulators) will start looking for buyers. A buyout offer for QC Ware from a larger player (possibly Honeywell or IonQ, looking to expand their software portfolio) is likely.

What to Do If You're an Investor

• QBTS (D-Wave): Hold with caution. The long-term thesis (quantum supremacy in optimization) has been undermined. But defense contracts with Anduril are real cash and protection. Selling on panic is a mistake, but adding to the position now is also not advisable. Wait for $15.

• Classical Semiconductors (NVIDIA, AMD): Upside. The industry is beginning to understand that for many 'quantum' problems, a good GPU and smart software are sufficient. NVIDIA especially benefits from CUDA optimizations for tensor networks.

• Avoid pure 'quantum optimization' startups without their own hardware. Their advantage is disappearing faster than expected.

Summary in one paragraph: What Joseph Tindall did on a laptop should make investors in quantum optimization companies question the basic thesis. D-Wave built a $7.5 billion business on the claim that spin glasses are quantum territory. The Flatiron Institute just showed that this is not the case. And they did it with algorithms that anyone can download from GitHub. The arms race between classical and quantum is just beginning, and the score is not yet in favor of quantum.

— Editorial Team