Why We Build at the Level of Atoms and Photons

Written By :

Category :

Company Views

Posted On :

Share This :

Every technology company rests on a bet about the world. Most of those bets are about markets. Ours is about physics, and it is worth stating plainly, because it explains nearly every decision we have made about what to build and how.

The bet is this: the universe computes, and the only machines that can keep up with it are machines made of the same things it is made of.

In the 1930s, Church and Turing proposed that anything that can be calculated by any effective procedure can be calculated by a Turing machine. That claim has never been refuted. In the decades that followed, complexity theorists strengthened it into a statement about efficiency: any reasonable physical model of computation can be simulated by a classical computer with at most polynomial overhead. This is the Extended Church–Turing Thesis, and it is the quiet assumption underneath all of classical computer science. It says the universe does not hide any computational power that a sufficiently large classical machine could not reproduce.

Quantum mechanics put that assumption under strain. Bernstein and Vazirani pointed out that the Turing machine is built on a classical picture of the world, while our best physical theory says the world is quantum. Shor’s algorithm made the strain concrete: a quantum computer can factor integers in polynomial time, and no known classical method can. The response was to revise the thesis rather than abandon it. The Quantum Extended Church–Turing Thesis states that a quantum Turing machine can efficiently simulate any realistic physical process.

Read carefully, that is an extraordinary claim. It says that the computational capacity of Nature is fully captured by machines built from Nature’s own substrate, and by nothing less. Any system evolving under the rules of quantum mechanics is sufficient to simulate anything the universe does efficiently. Atoms and photons are the purest such systems we know how to control. A classical machine is not. The gap between the two is not a matter of engineering effort. It is structural.

We take this proposition as the premise under which everything we build is engineered. Not because it has been proven, but because it has not been refuted, and because the only way to test it is to build machines that instantiate it.

A thesis of this kind cannot be settled by argument. It is a statement about what physical systems can and cannot do, and it is answered the way all such statements are answered: by constructing the systems and watching them.

This changes what a quantum machine is for. In the conventional framing, a quantum computer is an accelerator, a device that returns answers to certain problems faster than a classical computer could. That framing is not wrong, but it is narrow. If the thesis holds, a quantum machine is a controlled piece of the universe’s own substrate, prepared and measured under conditions we choose. It is closer to a telescope than to a processor. It lets us look at how Nature computes by setting up a region of Nature and asking it to compute.

Seen this way, a quantum computer, a quantum sensor, and a quantum network are not three product categories. They are three ways of putting the same substrate to work. A clock based on an atomic transition is a measurement of how Nature keeps time. An entangled link between two memories is a measurement of how Nature correlates distant events. Qubits evolving under an engineered Hamiltonian is a measurement of how Nature evolves a many-body state. The questions differ; the instrument is the same.

This is also why we are cautious about the economic story the industry tells about itself. Whether universal quantum computers deliver the commercial revolution attributed to them is an open question, and we would rather state that honestly than pretend otherwise. What is not open is that quantum systems, properly controlled, are the only instruments that can probe the quantum structure of the world directly. That is a sufficient reason to build them, and it is the reason we started.

If the thesis is the premise, coherence is the constraint.

Coherence is the condition under which a quantum system remains quantum: superpositions persist, phases stay defined, and interference between possibilities is still possible. Decoherence does not produce a wrong answer. It produces a classical one. A quantum system that loses coherence before a computation finishes has not refuted the thesis; it has simply stopped instantiating it and returned to the regime classical computers already describe.

Modern quantum information theory has formalized this. Coherence can be treated as a resource in the strict sense: there are states and operations that are free, in that they contain or generate none of it, and there are states and operations that cost something, in that they carry or preserve it. In the dynamical version of that theory, even the identity channel, the act of leaving a system alone, is not free, because every real system decoheres. Preserving coherence is itself a resource that must be paid for.

A logical qubit makes that cost tangible. Its coherence outlasts its physical constituents because error correction repeatedly extracts information about disturbances without revealing the state it protects. Like a refrigerator, this protection depends on removing entropy, not merely preventing its arrival. What appears to persist is continually sustained by fresh resources and corrective action. For us, this is what it means to treat coherence as a resource: not a property a device possesses, but a capability whose preservation must be built into the system.

This has a practical consequence that our industry has been slow to absorb. Coherence does not scale on its own. It is produced today in isolated systems, in quantities and under conditions suited to individual experiments. Every laboratory builds its own lasers, its own references, its own locking electronics, its own detection chains, and each of these is a local, hand-tuned source of coherence that cannot be moved, shared, or reproduced at will. The problems worth solving, both the scientific ones and the ones with commercial weight, are not bottlenecked by any single device. They are bottlenecked by the inability to generate, maintain, and distribute coherence across many devices at once.

So the engineering question that follows from the scientific premise is not “how do we build a bigger quantum computer.” It is “how do we make coherence a manufactured, distributable, resilient resource, rather than a laboratory condition.” That is what we mean when we say the mission is to distribute quantum coherence at industrial scale.

Any system that distributes coherence needs three things. It needs a shared notion of time precise enough that distant operations can be phase-coherent with one another. It needs a channel that can carry quantum states, not just classical descriptions of them, between locations without destroying them. And it needs the matter at the endpoints, the atoms that store, process, and measure quantum information, built to a standard that allows many such endpoints to exist and interoperate.

These are not features of a product line. They are preconditions imposed by the physics. A distributed quantum operation is only as coherent as the clock that references it, the link that connects it, and the memory that holds it. Weakness in any one collapses the whole into a set of independent classical devices.

Our choice to work in atoms and photons follows directly from this. Atoms are the best timekeepers Nature offers, and the most faithful quantum memories. Photons are the only carriers of quantum information that travel usefully far. The interface between them, light interacting with matter under precise control, is where time and information are constructed, and it is where the three preconditions meet. Everything we build sits at that interface.

There is a view in which modularity is simply good engineering practice. We hold a stronger view: modularity is what makes a thesis about Nature testable at scale.

A single, monolithic quantum apparatus can demonstrate a phenomenon once. It cannot tell you whether that phenomenon survives being reproduced, distributed, and connected. The classical computing world learned this with the transition from bespoke machines to standardized, composable components; the exponential progress that followed came not from any single breakthrough but from the ability to make the same unit many times and connect the units reliably.

We think the quantum world requires the same transition, and for a deeper reason. The thesis we are testing is about what networks of quantum systems can do, not what isolated ones can. Entanglement across a network, coherence across a fleet of clocks, computation across linked processors: these are the regimes where the quantum and classical pictures diverge most sharply, and they can only be reached by systems that are self-contained, manufacturable, and composable. A module that integrates coherent light, a stable reference, precise modulation, and quantum-limited detection into one unit is not a convenience. It is the smallest piece of the experiment that can be repeated.

It is sometimes asked how a company can pursue foundational questions about the computational structure of the universe while also delivering timing, networking, and sensing systems to defence and infrastructure customers. The honest answer is that the two are not in tension. They are the same work, at different stages of maturity.

A clock coherent enough to synchronize a distributed quantum computation is also a clock that holds its time when GPS is denied. A link that preserves entanglement across a fibre network is also a link whose integrity can be verified against interference at the level of physics rather than protocol. A module built to industrial reliability so that a research experiment can be repeated across a hundred sites is also a module that can be deployed in the field and trusted to run.

Treating coherence as a manufactured resource is what lets one capability serve both ends. The scientific program needs scale to ask its questions; the markets need resilience and reproducibility to trust the answers. We would not have been able to pursue the deeper questions without building instruments that are useful now, and we would not have built instruments worth deploying without being driven by questions larger than any single deployment.

We are often asked to predict when quantum technology will “arrive.” We decline, not from caution but from a conviction that the question is framed wrong. The universe has been computing quantum mechanically since the dawn of time. What is arriving is humanity’s ability to build instruments coherent enough, for long enough, across enough distance, to watch it do so and to put what we see to use.

To decode Nature’s foundational model, the instruments have to stay quantum long enough to ask the question. Distributing coherence is how we keep them there. Everything else follows.

error: Content is protected !!