Over the past decade, research has significantly advanced the science of quantum computing and led to the formation of many quantum startups. According to The Quantum Insider, the industry has grown to approximately 1,000 companies involved in some form of quantum technology. The long-term objective for quantum computing companies is to build large-scale, controllable, fault-tolerant quantum machines. However, that is a complicated process rife with difficult engineering and physics challenges that are yet to be solved.
I recently had the opportunity to talk to Yuval Boger, CMO for QuEra Computing. It had been a while since our last conversation, and I was looking forward to hearing about QuEra’s progress and its latest research efforts.
QuEra began as a startup in 2018 using technology developed by MIT and Harvard researchers. The company uses neutral-atom qubits for its Aquila quantum computer, which runs on a field-programmable qubit array (FPQA) processor, with up to 256 rubidium atoms for qubits. The FPQA architecture allows qubit configurations to be rearranged on demand without the need to change the hardware, which means that it could also be called a software-defined quantum computer. One of FPQA’s other unique features is that it can operate in a dual analog and digital mode.
Newly expanded access for customer
QuEra recently expanded how customers can use its quantum service. Since November 2022, customers have been able to access QuEra’s system through Amazon Braket, a fully managed quantum computing cloud service designed for quantum computing research and software development.
At the beginning of this month, the company announced that customers can also use its quantum machines directly through QuEra’s Premium Access service. According to Boger, the new access method was created based on requests from QuEra customers. “Customers, including a national lab, have been asking for direct access to our system,” Boger said. “While Braket is a great service, customers sometimes need the ability to work directly with our scientists. Basically, customers felt they could accomplish more by having direct communications with QuEra.”
Forbes Daily: Get our best stories, exclusive reporting and essential analysis of the day’s news in your inbox every weekday.
In the press release announcing these new options, QuEra CEO Alex Keesling had this to say: “As we ramp up the production capabilities and expand our exceptional team of application-focused scientists, we’re thrilled to unlock additional avenues for engaging with our ground-breaking technology. The launch of our on-premise and premium access models stems directly from resonant customer demand. This pivotal move is not just a response but an exciting leap forward that opens a realm of new opportunities for our customers and for QuEra.”
Although Premium Access costs more than the Braket option, Boger added that it is already a popular offering for many of QuEra’s customers. Boger also noted that as part of these offerings, QuEra can now provide clients with not only secure remote access, but also higher service level agreements and a reservation system that allows researchers to reserve machine time to avoid waiting for a turn.
QuEra also introduced a leasing option so that customers can have a QuEra quantum computer on-premises rather than accessing it remotely through the cloud.
“We are seeing an explosion of interest in national, regional and corporate users that want a quantum computer on site,” Boger said. “It could be for various reasons such as national pride or a large defense contractor that doesn’t trust anything on the cloud for security reasons and needs an air-gapped system. It could also be someone that just wants full control and doesn’t want to wait in queue behind a large company with large jobs.”
Digital vs. analog
Today’s quantum computers use a variety of architectures and technologies to create basic quantum computation units called qubits. Common physical implementations of qubits include photons, atoms, ions trapped in electromagnetic fields and manufactured superconducting devices. The choice of qubits dictates operational factors such as temperature, type of control, applications and scalability.
Most well-known quantum computer companies use digital gate-based architectures and logic gates within circuits to control the quantum state of qubits. Here are a few companies that use gates: Atom Computing (neutral atoms), IBM (transmon superconductors) and IonQ and Quantinuum (trapped ions).
QuEra’s Aquila is not a gate-based quantum computer, at least not yet. QuEra’s machine is classified as an analog quantum computer because its qubits are manipulated by gradually fine-tuning the states.
QuEra’s qubits are created from rubidium-87 atoms by using the electron in the outer shell of each atom to encode quantum information. The electron can exist in a combination of two spin states that represent the 0 and 1 states of a qubit. QuEra’s analog mode works well for optimizations, modeling quantum systems and machine learning.
QuEra’s hardware is complemented by Bloqade, the company’s open-source software development kit, which allows users to design, simulate and then execute programs. In more precise terms, Bloqade is an emulator for the Hamiltonian dynamics of neutral-atom quantum computers.
You can’t discuss QuEra’s analog quantum computer without talking explicitly about Rydberg states. These atomic states play a major role in QuEra’s architecture and deserve a bit more scientific explanation.
Rydberg states are created by boosting rubidium-87’s single valence electron to a very high energy level. Electrons normally orbit the nucleus at low energy levels and near the nucleus. But the outer electron in Rydberg atoms has an artificially-induced orbital radius that is sometimes thousands of times larger than normal. Because of Rydberg atoms’ large size and the distance between the outer electron and its nucleus, these atoms possess exaggerated properties that make them very sensitive to electric and magnetic fields.
QuEra uses Rydberg atoms’ outer electrons to create two qubit states. The interaction distance between Rydberg atoms allows a form of conditional logic. Atoms far apart can act independently, while atoms close to each other allow only one Rydberg excitation to occur. This limiting effect is called a Rydberg blockade. The point of all of this for QuEra is that flexible geometries and Rydberg blockades, guided by laser tuning controls, can be used to implement quantum algorithms.
In summary, QuEra’s neutral atoms provide reconfigurable and controllable qubits, and their interactions can create conditional logic. These features can be used for quantum simulation and optimization in ways that can’t be achieved with hardware qubits.
QueEra has developed a method to shuttle atoms to different locations while still maintaining its quantum states. Shuttling allows connectivity between the rubidium atoms to be reconfigured as needed to handle complex problems.
Three zones are involved in shuttling: the memory zone, where lower energy states with longer coherence are stored; the processing zone, where operations take place; and the measurement zone, where qubits can be isolated and read without disturbing other qubits.
Boger gave me a simple explanation of the zones that also suggests how QuEra’s next generation will handle qubit operations. “If you have these three zones, you don’t need 10,000 control lines for 10,000 qubits,” he said. “You can shuttle the qubits into the compute zone, then run it. Do the operation, then take it from there. It’s simple.”
QuEra’s shuttling is similar in concept and function to Quantinuum’s QCCD trapped-ion architecture. QuEra has also developed a fast transport method optimized to avoid motional heating of the atoms, which can cause a loss of fidelity in the shuttling process.
If quantum computing is to fulfill its potential, we will need the capability to scale qubits into the millions. Of course, error correction will play a major role in reaching that number. Currently, the maximum number of qubits in use is around 500. But a number of companies, including QuEra, are expected to announce much more than that sometime soon.
When the issue of scaling came up during my discussion with Boger, I was surprised by how far QuEra had come. He showed me an image of 10,000 laser spots that can contain 10,000 atoms in a 100 x 100 array.
“Considering our current capabilities,” he said, “we believe we can get to at least 10,000 qubits without needing interconnects. The 10,000 laser spots on this image were created by the optical tweezers used to capture the atoms. Each atom is only three or four microns apart. It is also an advantage that our qubits function without cryogenic cooling.”
Seeing so many qubits in such a small area was impressive. Even so, to put 10,000 qubits into production will require error correction or, at a minimum, extremely effcient error mitigation.
The good news is that analog quantum computers require less error correction than digital gate-based machines. Still, putting such a high number of qubits into operation would also require higher qubit fidelities than possible today, even though QuEra’s collaborators at Harvard obtained a two-qubit gate fidelity of 99.5%.
Scaling a quantum computer to that level will require a great deal of clever physics, along with the precise engineering to put it into practice.
QuEra has also done some research with analog quantum simulation of topological matter. Without going into too much technical detail, topological matter refers to a class of quantum materials that have unique properties derived from their underlying topology. Among other benefits, topological matter can be resistant to noise, which could also make it useful for error resistance.
The existence of topological material was predicted theoretically more than five decades ago. It has taken fifty years just to determine that it actually exists—which should be an indication of how technically challenging it is going to be for anyone to develop topological qubits.
QuEra isn’t alone in researching the topic. Google published a paper in late 2021 describing the creation of topological ordered states using semiconductors. Earlier this year, Quantinuum announced a topological discovery of its own; the company has a full program dedicated to this research. After a rocky start a few years ago, Microsoft has re-announced its intentions to build a quantum computer using a hardware form of Majorana topological qubits.
Creating a useful topological quantum computer is likely to be ten or more years away, but I will be following topological advancements as they are made.
Maximum Independent Set (MIS)
QueEra’s optically trapped neutral atoms allow flexibility in qubit arrangements. Unlike in microchips, optical tweezers can position the atoms into any geometric 2-D position. Their arrangement relative to each other determines how the qubits interact—a key factor in quantum computing. Furthermore, tweezer control allows the connections to be dynamically reconfigured, which can alter properties of the quantum processor.
These advantages enable QuEra’s 256-qubit quantum computer to use a unique method of solving optimization problems of the Maximum Independent Set (MIS) type. An MIS problem can be solved by mapping the geometry of the problem, such as the geographic layout of radio antennas, directly into the hardware. Many industrial problems are constrained by physical layouts, making them candidates for being solved as a MIS. There are a number of areas where MIS can be useful:
- Resource allocation, e.g., finding the maximum number of tasks that can be scheduled simultaneously when the tasks have conflicting resource requirements
- Social network analysis, e.g., identifying the most influential people in a social network who are not directly connected to each other
- Map problems, e.g., locating radio antennas at optimum sites without excess overlapping broadcast areas
- Pattern detection, e.g., finding anticorrelated elements in a network, such as suppliers in a supply chain
QuEra sees one of its future challenges as moving beyond small proofs-of-concept to larger quantum systems that demonstrate higher values and more impacts sooner. To support this, on top of using its FPQA and analog approach, QuEra has implemented hybrid quantum-classical algorithms for solving relevant problems.
One such demonstration optimized placements of gas stations across city locations by encoding the geometry of the problem into qubit positions, then measuring the system’s ground state. The hybrid approach found solutions comparable to or slightly better than classical algorithms alone. While this is not definitive quantum advantage, it does indicate the feasibility of testing quantum optimization algorithms on actual quantum hardware.
Even though analog quantum computers can’t ever match the capabilities of a universal gate-based quantum machines, there is a place for analog technology in the areas of simulation, optimization and machine learning. QuEra’s approach will be differentiated by the use of FPQAs to allow flexible encoding of problems directly into the qubit geometry.
Over a relatively short time, QuEra has assembled experts in the areas of chip-scale photonics, ultra-stable lasers and precision control systems. It has expertise in all the required areas of software, applications and algorithms needed to be successful in quantum. QuEra has over 50 employees working in the areas of hardware, software and business operations, and its MIT and Harvard heritage is a major asset for continued technical advancement.
QuEra’s neutral-atom analog quantum computer provides some capabilities unavailable with classical computers. However, it is not yet close to the technical requirements needed for a fault-tolerant quantum computer capable of solving world-class problems such as drug design or climate change. Currently, all quantum computers, whether analog or digital, still have technical problems to overcome in the areas of fidelity, scale and full error correction.
QuEra has identified its major sources of errors and it is working to reduce them. These include laser noise, atom motion, state decoherence and scattering, imperfect laser functioning and measurement errors.
After QuEra converts its architecture to a digital mode, there are several challenges that must be overcome before fault-tolerance becomes possible. Beyond large numbers of qubits and a high two-qubit gate fidelity, we don’t yet know what ratio will be needed between physical qubits and logical qubits. It will likely vary depending on which qubit technology is used. Google has done extensive work on error correction, scaling between 17 and 49 physical qubits per logical qubit. It believes, as QuEra does, that it will be possible to use logical qubits to build a large-scale error-corrected quantum computer.
QuEra’s future research will be directed at increasing the number of qubits, operation fidelities and levels of connectivity. My perspective is that QuEra is pushing the boundaries of analog quantum computing—and that its technology warrants attention. The company has a flexible architecture and intriguing capabilities, and its customers’ steady demands for easier and closer contact methods is reason enough to be optimistic about QuEra’s traction in the market.
However, QuEra and the entire industry face an immense technical challenge to raise quantum computing to its true potential. Achieving quantum advantage would be a great half-step and a signal that fault-tolerance is only a few years away.