Building the Computer

At Oratomic, our mission is to build the world's first utility-scale fault-tolerant quantum computers.

The universe runs on quantum mechanics; classical physics is only an approximation. Some things can never be fully understood classically, and it seems that the ultimate form of computation and intelligence cannot be purely classical. Quantum computing offers a fundamentally new model of computation that could deepen our understanding of the universe and have consequences far beyond what we can predict today.

The central challenge is that quantum computers are extremely sensitive to noise. When Peter Shor invented his algorithm in 1994, showing the first clear evidence that quantum computers could solve important problems far beyond a classical computer, the reaction was serious skepticism. With qubits in superpositions between 0 and 1, quantum computers reminded people of analog computers, which, while theoretically more powerful than digital computers, work directly with continuous values, so small errors compound exponentially and eventually derail a computation. Digital computers are predominantly used because bits are robust and make error correction possible. Since quantum states are continuous, quantum computing seemed destined for the same fate, and later theoretical work confirmed this: without error correction, decoherence effects render large-scale quantum computation effectively classical, and quantum advantage disappears.

The breakthrough came in 1995 with the invention of quantum error correction, which showed how to encode an analog-type computation on a logical qubit while realizing digital-type correction on the underlying physical qubits, a kind of wave-particle duality for computation itself. This analog-type computation with digital-type correction is only possible because of quantum mechanics, and it is the key feature that makes quantum computing a genuinely new computational paradigm.

But quantum error correction is enormously demanding. Conventional wisdom has held that a fault-tolerant quantum computer capable of large-scale useful applications requires millions of qubits. While a quantum industry has existed for some time, it has centered around uncorrected quantum computers, despite strong theoretical evidence that error correction is required for useful applications (and as such, no actual commercial utility has been achieved yet). Compounding the difficulty, quantum computers have historically controlled each qubit individually, creating the so-called "wire problem": scaling to millions of qubits seemed to require millions of individual control lines. Altogether, building such a device has long seemed far out of reach.

Over the past several years, Oratomic's founding team, alongside other researchers, has worked to develop a new approach to fault-tolerant quantum computing based on reconfigurable arrays of atoms trapped in focused laser beams. Physical qubits are encoded in individual atoms, which are moved and entangled with each other by global laser pulses to implement error-corrected quantum computations.

This approach has three key advantages:

(1) Parallel, efficient classical control. Because every atom is identical, a single global laser pulse can apply the same operation to many qubits at once — enough to implement the redundant operations error correction requires. In principle, tens of thousands, even millions, of qubits can be controlled with only several control lines.

(2) Massive, parallel nonlocal connectivity. Atomic qubits can be reconfigured in parallel to form nonlocal connections, driving major advances in quantum error correction that reduce the space-time overhead of computation — and with it the resources needed to run the computer — by a factor of a thousand or more.

(3) Many qubits at low cost. Identical neutral particles are plentiful and simple to control: atomic physicists have spent decades laser-cooling them for quantum applications, and systems with hundreds of millions of coherent atoms are routinely handled with room-temperature apparatuses costing only a few million dollars. Controlling individual atoms is harder, but the underlying physics allows for controlling even millions of qubits on a tabletop system. Coupled with the ability to understand and predict atomic errors from first principles, these systems are intrinsically scalable and engineerable.

Even with these advances, Oratomic's founding team did not originally think a quantum computer was close to being commercially useful. That changed with a recent finding: because of the nonlocal connectivity of this architecture, an error-corrected computer capable of running Shor's algorithm may need as few as ten to twenty thousand atomic qubits, and systems trapping comparable numbers of atoms already exist in the lab.

Building this computer will be a formidable engineering challenge, but much of the required performance has already been demonstrated, and with significant room for improvement beyond the current state of the art. The largest coherent system sizes, at scales comparable to what the computer will require; the highest qubit fidelities, at the performance needed for fault-tolerant operation; and the most advanced error-corrected algorithms, demonstrating a full architecture for universal, fault-tolerant computation — all of these state-of-the-art results have been realized in university labs, in apparatuses that weren't even designed for fault-tolerant quantum computing. Only very recently have we crossed the threshold where all the components have reached required performance, and one can devise a coherent plan for a fault-tolerant quantum computer.

How much of the engineering ahead is "quantum"? In some sense, very little. Quantum error-correcting codes can be understood as built from two classical codes — one correcting bit errors, one correcting phase errors — and error-correction performance is well modeled by efficient classical simulations. We’ve previously run many experiments confirming that the atomic qubits reproduce these simulations and further experiments tying atomic performance directly to the precision and intensity of laser light. While working large-scale error-correction will be an extraordinary physics milestone, from an engineering perspective it mainly confirms that the atoms match the simulations and that the classical decoding algorithm achieves exponentially low error. The “truly quantum” part is the quantum algorithm that runs on the logical qubits, but nearly all the engineering required to get there is classical: advanced optics, electronics, atomic physics, mechanical engineering, controls, software, algorithms, and artificial intelligence, all working together at scale and with high precision.

Building this fault-tolerant quantum computer will be one of our generation's defining technological achievements, but we believe it can be done best by a small, focused team. At Oratomic, scientists, engineers, and technicians work closely across disciplines because the machine cannot be cleanly divided into independent problems: decisions in optics affect controls, mechanics affect stability, electronics affect the atomic system, and improvements in one part of the machine can change what's possible elsewhere. People develop a broad understanding of the system, and individual contributions can shape the computer as a whole. A background in quantum computing is not required. We're looking for people who are exceptional at what they do, who like working closely with others on hard problems, and who want to build something that has not existed before.