Key Takeaways
- In 2025, the quantum computing field moved beyond one-time claims of supremacy toward a framework of quantum advantage, emphasizing reproducible, testable results that outperform classical systems in specific domains.
- Google’s 2025 experiment achieved a breakthrough in quantum computation by reversing quantum chaos to observe constructive interference, completing in two hours what would take classical supercomputers three years.
- The future of quantum computing lies in hybrid systems where quantum processors act as specialized accelerators within supercomputing centers, working alongside CPUs and GPUs to tackle problems in chemistry, logistics and physics.
For nearly a decade, "quantum supremacy" was the moon landing of computing, a bold claim that a quantum processor had done something no classical computer could. The term came from Google's 2019 feat, when a 53-qubit chip solved a random sampling task faster than the world's largest supercomputer.1 Yet the triumph was short-lived. Within months, IBM and others found smarter ways to simulate the same task, closing the gap.2 What should have been a clean break became a tug-of-war between evolving quantum hardware and ever-stronger classical software. By 2025, the field had matured past that binary notion of victory. The new goal, quantum advantage, recognizes progress as domain-specific, testable and dynamic, a frontier that moves rather than a flag planted once.
Redefining the Finish Line
In 2025, IBM and PASQAL reframed the quantum race with a more scientific mindset. Their paper, "A Framework for Quantum Advantage,"3 proposed that progress shouldn't hinge on one flashy claim of victory but on credibility and repeatable results, the way we now treat AI model benchmarks.
They outlined two tests. First, validation: Can we trust the output? Quantum results need error bars,4 reproducibility and cross-checks against classical methods, just as AI results require transparent datasets and evaluation metrics. Second, quantum separation: Does the quantum machine actually outperform its classical rivals, faster, cheaper or more accurately, much like an AI model beating the previous state-of-the-art?
This isn't a race to a finish line but an evolving leaderboard. Each experiment is a hypothesis to test, not a trophy to defend. If classical computing catches up, it's not a defeat; it's progress for the entire field. In this light, quantum advantage isn't about one machine conquering another; it's about building cumulative, verifiable trust in a new form of computation.
Where Quantum Advantage Might First Emerge
If "quantum advantage" is the new finish line, the obvious question is: Where might we actually cross it first? IBM's framework offers three broad categories of problems where quantum computers could begin to outperform classical ones, not in theory, but in practice. Each of them sounds abstract until you look at what it means in real life.5
The first category is sampling problems, exercises in controlled randomness. Imagine a logistics firm trying to optimize thousands of delivery routes while accounting for weather, traffic and fuel. A classical computer can only approximate the best answer, but complexity grows explosively. Quantum computers thrive in that chaos. Instead of estimating outcomes bit by bit, they generate and analyze massive probability distributions directly, uncovering efficient patterns hidden in the noise. From financial portfolios to molecular design, quantum sampling could reveal solutions that classical systems can only glimpse.
The second family, variational problems, sits closer to chemistry than code. Imagine finding the most stable electron arrangement in a new material, key to better batteries or catalysts. Classical supercomputers can only approximate these quantum interactions; the math grows too fast. Variational quantum algorithms tackle this as a search game: The quantum processor suggests configurations, the classical computer scores them, and together they refine toward the lowest energy state. The result isn't just a number; it's a potential molecule or material that could power real technologies.
The third path, expectation-value problems, may sound abstract, but it forms the backbone of physics. These values capture what happens on average in quantum systems, like a material's magnetic moment or a molecule's energy. Accurate measurement drives fields from drug design to semiconductor research. The obstacle is noise: Real qubits blur these faint signals. New error-mitigation methods, which mathematically cancel out that static, now let quantum devices estimate such averages with precision rivaling top classical simulations.
Taken together, these three problem families sketch a road map for where early, credible quantum advantage could appear:
- Sampling could transform optimization, logistics and cryptography.
- Variational approaches could accelerate chemistry and materials breakthroughs.
- Expectation values could deepen our ability to model and predict the physical world.
Each path is still experimental, but they all point toward something bigger: quantum computers beginning to do work that doesn't just impress physicists, but matters to the rest of us, shaping how we make things, move things and understand the universe itself.
The Quiet Revolution beneath the Quantum Horizon
For most people, quantum hardware sounds like science fiction, superconducting chips and subatomic magic. Yet the real revolution isn't happening apart from AI or GPU data centers; it's happening inside them.
IBM and PASQAL call this the quantum-centric supercomputing era.6 The idea is simple: Quantum processors won't replace classical machines but work beside them, as GPUs once did for AI. Just as GPUs turned deep learning from a niche into a trillion-dollar industry, quantum processing units (QPUs) are becoming specialized accelerators within high-performance computing clusters.
Picture the division of labor. CPUs manage logic and data flow. GPUs handle massive numerical workloads for AI. But problems like molecular simulation or quantum-material modeling involve superposition and entanglement, nature's own complexity. That's where QPUs shine, directly modeling these interactions while CPUs and GPUs manage the rest.
IBM's Heron processor, with 156 qubits, can execute 250,000 circuit layers per second while correcting errors in real time, a built-in quality-control loop. PASQAL takes a different path: using neutral atoms trapped by lasers, arranged like 3D chess pieces to act as analog quantum simulators, soon scaling to 250 qubits.
Both systems are already being installed in national supercomputing centers such as GENCI in France and Jülich in Germany, running under the same job schedulers scientists use today. Soon, researchers will submit hybrid workloads, part classical, part quantum, as easily as GPU jobs.
The next era won't be a quantum takeover but a quantum alliance, where CPUs, GPUs and QPUs share the same digital bloodstream.
At the Edge of Ergodicity:7 Google's Quantum Mirror
For years, quantum research has been like learning an alien alphabet: intricate symbols, strange rules, endless practice. Then, in 2025, Google finally wrote a sentence. Its paper in Nature, titled "Observation of Constructive Interference at the Edge of Quantum Ergodicity,"8 sounds arcane, but behind the jargon is a surprisingly human story: what happens when information is thrown into chaos, and how we're learning, astonishingly, to watch it come back.
Picture shouting into a canyon so deep that your voice never echoes; it dissolves into countless tiny reflections. That's what happens in a quantum many-body system. When particles interact, their information—their spin, position and energy—spreads and scrambles so completely that it seems lost forever. Physicists call this quantum chaos, and for decades they've wondered: when information disappears, is it truly gone or just hidden?
Google set out to find out. Using 65 superconducting qubits, its team built a kind of quantum mirror. They pushed the system to the edge of chaos, then did something extraordinary: They reversed time. They programmed the qubits to unwind their evolution, step by step, searching for faint traces of order that might still linger. To their surprise, they found them.
What they saw, a phenomenon called constructive interference, was the quantum equivalent of faint echoes merging and amplifying, briefly revealing patterns inside the noise. It's as if, after shouting into the canyon, a coherent whisper returned, carrying traces of your original voice. This effect, once thought unobservable, emerged directly from the data.
The experiment's beauty lay not just in the physics but in its computational magnitude. Simulating the same process on the world's fastest classical supercomputer, Frontier, would have taken three years; Google's quantum chip did it in two hours. And, unlike early "supremacy" claims, this result wasn't a one-off stunt. It cleared the most rigorous benchmark of scientific reliability, a signal-to-noise ratio greater than two, verified and reproducible.
From Trick to Tool
The leap from Google's 2019 "quantum supremacy" experiment to this moment mirrors the evolution of AI from party trick to powerful tool. The 2019 Sycamore experiment was like a magician's shuffle, technically dazzling, but with no direct application. The 2025 study is closer to a microscope: a precision instrument that lets scientists observe the unseen dynamics of nature.
Those exotic correlators aren't just math for its own sake. They describe how energy and information flow through matter, insights that help explain superconductivity, catalytic reactions and even how quantum coherence might persist in biological systems. Google's machine didn't just observe this behavior; it learned from it. By comparing experimental data with simulated models, it inferred hidden constants, the physical "rules" governing a system, faster than any known classical approach. This technique, known as Hamiltonian learning, transforms the quantum computer into an experimental physicist in its own right.
The New Currency: Trust
Across the field, a quiet consensus is forming: Trust is the new performance metric. It's no longer about who has the most qubits, but who has the most verifiable results. Quantum progress now hinges on three questions:
- Validation: Are the results reproducible and backed by quantifiable error bars?
- Complexity Separation: Can classical algorithms match them within a reasonable time?
- Relevance: Do they reveal something useful about the real world?
Google's 2025 work answers yes to all three. The experiment bridges theory and observation, showing that quantum processors can do more than compute; they can discover.
If 2019 proved that quantum computers could run, 2025 proved they could matter.
1 Source: F. Arute, K. Arya, R. Babbush, et al. "Quantum supremacy using a programmable superconducting processor," Nature 574 no. 7779 (2019): 505–510.
2 Source: E. Pednault, J. A. Gunnels, G. Nannicini, L. Horesh & J. M. Gambetta, "On ‘Quantum Supremacy'" [technical blog post], IBM Research Blog, 10/21/19.
3 Source: O. Lanes, M. Beji, A. D. Córcoles, et al., "A framework for quantum advantage," arXiv preprint, 2025. arXiv:2506.20658.
4 Error bars quantify the uncertainty in the measured expectation value or computed observable that the quantum processor outputs. They tell you how trustworthy that reported result really is.
5 Source for the three types of problems: Lanes et al., 2025.
6 Source for the hardware section: Lanes et al., 2025.
7 Ergodicity means that, over time, a system explores all its possible states so thoroughly that time averages equal overall averages — randomness becomes predictably distributed across the system.
8 Source: Google Quantum AI and Collaborators. (2025). Observation of constructive interference at the edge of quantum ergodicity. Nature, 646(7994), 825–831.
A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.
U.S. investors only: Click here to obtain a WisdomTree ETF prospectus which contains investment objectives, risks, charges, expenses, and other information; read and consider carefully before investing.
There are risks involved with investing, including possible loss of principal. Foreign investing involves currency, political and economic risk. Funds focusing on a single country, sector and/or funds that emphasize investments in smaller companies may experience greater price volatility. Investments in emerging markets, currency, fixed income and alternative investments include additional risks. Please see prospectus for discussion of risks.
Past performance is not indicative of future results. This material contains the opinions of the author, which are subject to change, and should not to be considered or interpreted as a recommendation to participate in any particular trading strategy, or deemed to be an offer or sale of any investment product and it should not be relied on as such. There is no guarantee that any strategies discussed will work under all market conditions. This material represents an assessment of the market environment at a specific time and is not intended to be a forecast of future events or a guarantee of future results. This material should not be relied upon as research or investment advice regarding any security in particular. The user of this information assumes the entire risk of any use made of the information provided herein. Neither WisdomTree nor its affiliates, nor Foreside Fund Services, LLC, or its affiliates provide tax or legal advice. Investors seeking tax or legal advice should consult their tax or legal advisor. Unless expressly stated otherwise the opinions, interpretations or findings expressed herein do not necessarily represent the views of WisdomTree or any of its affiliates.
The MSCI information may only be used for your internal use, may not be reproduced or re-disseminated in any form and may not be used as a basis for or component of any financial instruments or products or indexes. None of the MSCI information is intended to constitute investment advice or a recommendation to make (or refrain from making) any kind of investment decision and may not be relied on as such. Historical data and analysis should not be taken as an indication or guarantee of any future performance analysis, forecast or prediction. The MSCI information is provided on an “as is” basis and the user of this information assumes the entire risk of any use made of this information. MSCI, each of its affiliates and each entity involved in compiling, computing or creating any MSCI information (collectively, the “MSCI Parties”) expressly disclaims all warranties. With respect to this information, in no event shall any MSCI Party have any liability for any direct, indirect, special, incidental, punitive, consequential (including loss profits) or any other damages (www.msci.com)
Alejandro Saltiel, Andrew Okrongly, Behnood Noei, Bradley Krom, Brendan Loftus, Brian Manby, Christopher Gannatti, David Graichen, Hyun Ku Kang, Jeff Weniger, Jeremy Schwartz, Jonathan Steinberg, Joseph Grogan, Joseph Tenaglia, Kara Dombroski, Kevin Flanagan, Lauren Pfendt, Liqian Ren, Lonnie Jacobs, Matt Wagner, Rick Harper, Ryan Krystopowicz, and Vanya Sharma are registered representatives of Foreside Fund Services, LLC.
WisdomTree Funds are distributed by Foreside Fund Services, LLC, in the U.S. only.
You cannot invest directly in an index.
© WisdomTree, Inc.
Read more commentaries by WisdomTree, Inc.
