Social media has erupted over the past few months with grand announcements from Google Quantum AI, then Microsoft, and now AWS. Unfortunately, like an archaeological dig, the valuable artifacts of knowledge we’re looking for have been buried under layers of hype, misinformation, and jargon. These artifacts have been recovered, however, and are on display for all to see.
Google Quantum AI’s “Willow”
The Announcement
In December 2024, Google introduced its Willow quantum chip [1], which represents a major advancement in quantum computing. Willow significantly reduces errors as it scales, addressing a long-standing challenge in quantum error correction. The announcement also highlighted that the chip can perform certain computations in under five minutes that would take a supercomputer 10 septillion years, emphasizing its capacity to solve complex problems beyond the reach of classical computing.
What does that mean?
The key point here is Willow's ability to achieve "below threshold" error rates while increasing qubits. This breakthrough marked a crucial step towards commercially viable quantum computers. Let’s explain what that means.
The “Willow” chip accomplished several things. First, it doubled the number of physical qubits that were previously available on a Google Quantum AI chip. Second, without getting technical, physical qubits can retain information for a short period of time, and this duration was multiplied by five. This duration, called “coherence,” provides a window of opportunity for computation. Computation must be completed within this window, and this window is now roughly five times longer than it used to be, allowing roughly five times as many operations during the execution of an algorithm. Third, the accuracy, or “fidelity,” of performing operations improved. [2]
Most importantly, “Willow” demonstrated that “logical qubits” can improve coherence further. The crucial point here is that using more physical qubits, extend coherence as we predict it should. We can add more physical qubits to our logical qubits, thus opening up the “window” long enough to faithfully execute our desired algorithms. [2]
Microsoft’s “Majorana 1”
The Announcement
On February 19, Microsoft introduced the Majorana 1 [3], a quantum computing chip based on “topological” qubits. This advancement marks a significant step in quantum technology, aiming to develop a scalable, million-qubit processor that addresses complex industrial challenges. Majorana-based qubits promise to reduce measurement, control, and error correction issues prevalent in other qubit types.
Skepticism
At the time of this writing, there is insufficient evidence to prove that its “Majorana 1” chip is what they claim it to be. Skepticism is heightened by the knowledge that Microsoft has previously retracted a related paper. However, the excitement stems from the possibility of topological qubits finally becoming a reality.
What does that mean?
Whereas Willow’s key breakthrough adds physical qubits to create better logical qubits, topological qubits can be thought of as being protected by physics itself. The potential benefits include enhanced scalability, (up to a million qubits on a single processor), reliability, (reduced need for extensive error correction), and the capability to solve industrial-scale problems more efficiently. Instead of needing millions or even billions of physical qubits to do what we want to do, the number of naturally-resilient topological qubits we would need is substantially less. [4]
If Microsoft’s claims withstand peer review, the modality is decades behind others. However, that could be short-lived depending on how robust they turn out to be. Speculatively, they could shoot right past the other modalities, analogous to transistors overtaking vacuum tubes. Again, speculatively, that would change the projected timelines for commercially-useful quantum computing. What’s interesting is that Microsoft is pointing to the age of the paper that is currently in question. Although the paper has not satisfied peer review, Microsoft’s public statements are reportedly based on more recent, yet-to-be-published data. Again, there is skepticism based on the historical retraction. However, there is a combination of optimism and hope that this unpublished data will satisfy peer review, because that would indeed be exciting. [4]
AWS’s “Ocelot”
The Announcement
AWS has announced Ocelot, a new quantum computing chip developed by the AWS Center for Quantum Computing at Caltech. Ocelot significantly reduces the costs of quantum error correction by up to 90% compared to current methods, marking a breakthrough towards fault-tolerant quantum computers. The chip uses "cat qubits," which inherently suppress certain errors, and integrates error correction directly into its architecture. This approach could make quantum computers smaller, more reliable, and cheaper, accelerating the timeline for practical quantum computers by up to five years.
What does that mean?
The “Ocelot” chip can be thought of in some sense as being somewhere in-between “Willow” and “Majorana 1.” This type of qubit, called a “cat qubit,” is in the same family of “superconducting” qubits as Willow’s “transmon” qubits, but they are engineered to have lower error rates. An explanation would get rather complicated, so the key takeaway is fewer errors. Cat qubits still need to be encoded as logical qubits, but substantially fewer cat qubits are needed per logical qubit to achieve the same error rates as compared to transmons. [5] Therefore, to perform comparable computation, we would need the fewest topological qubits, more cat qubits, and even more transmon qubits.
Like Google Quantum AI, AWS demonstrated that larger error correction codes produce stronger benefits. However, AWS demonstrated that it could implement comparable codes using noticeably fewer physical qubits. As the codes get larger, the savings in regard to physical qubits will become even more pronounced. [6] It is important to note that there are differences in coherence times and various error rates, but the key takeaway is that cat qubits can detect and correct errors with fewer total physical qubits compared to transmons. Commercially-useful algorithms will run on relatively small cat qubit quantum computers compared to their transmon counterparts.
Conclusion
Google Quantum AI and AWS advancements in error correction, have marked a crucial step towards commercially viable quantum computers.
The announcement of Microsoft's Majorana 1 chip is relevant as it signifies a notable technological advancement. However, it is unlikely to significantly shorten the timelines for practical, useful quantum computing due to existing challenges. While there is genuine progress, media coverage has tended to be overly hyped. This development strengthens Microsoft's position but does not definitively place them at the forefront of quantum computing, given the strong competition in the field.
Even though the hardware isn't fully developed yet, it's now a question of when, not if. Additionally, quantum technology isn't plug and play. Thus, the financial sector should proactively work on understanding how to integrate quantum computing into their operations to be prepared when commercially viable quantum computers become available.
Zenith Emerging Technologies SIG Discussion
References
- Google -"Meet Willow, our state-of-the-art quantum chip", December 9, 2024.
- Shtetl-Optimizer, the blog of Scott Aaronson - "The Google Willow thing", December 10, 2024.
- Microsoft - "Microsoft’s Majorana 1 chip carves new path for quantum computing", February 19, 2025
- Shtetl-Optimizer, the blog of Scott Aaronson - "FAQ on Microsoft’s topological qubit thing", February 20, 2025.
- Amazon - "Amazon Web Services announces a new quantum computing chip", February 27, 2025
- Amazon Science - "Amazon announces Ocelot quantum chip", February 27, 2025
Author: Carmen Recio, Director Quantum Computing at Moody's