Imagine arranging single atoms like pixels — nudging individual atoms into a glowing grid with beams of light, in any pattern you like — and, in the lab next door, borrowing the very transistor factory that etches billions of components onto a chip and using it to stamp out qubits instead. These are not thought experiments: they are two of the fastest-rising quantum-hardware platforms, neutral atoms and silicon spin qubits. Both are relative newcomers next to superconducting circuits and trapped ions, and both are climbing fast — for reasons worth understanding.
Every candidate technology has to answer the same checklist — the
A neutral atom qubit is exactly that — a single, uncharged atom (often rubidium or a
strontium/ytterbium species), with the qubit stored in two of its long-lived internal states, say a
ground level
Two features make this exciting. First, the atoms are identical by nature — every rubidium atom is a perfect copy of every other, so there is no fabrication spread — and you can already assemble hundreds to thousands of them. Second, because the traps are just light, the geometry is flexible and reconfigurable: arbitrary 2D or 3D layouts, and atoms can even be physically rearranged mid-circuit, shuttling qubits next to whichever partners they must interact with.
Neutral atoms are, well, neutral — they normally ignore each other, which is wonderful for keeping a qubit quiet but useless for a two-qubit gate. The entangling trick is to briefly promote an atom to a Rydberg state: an enormously puffed-up state with one electron kicked to a very high orbit. A Rydberg atom is thousands of times larger than normal and interacts strongly with any other nearby Rydberg atom.
That strong interaction produces the Rydberg blockade. If one atom is already excited to its Rydberg state, it shifts the energy of its neighbours so far that the laser meant to excite them no longer matches — so within a certain blockade radius, a second atom simply cannot also be excited. One excitation locks out the rest. Step through the figure to watch it happen.
Put two atoms, a control and a target, close enough to sit inside one blockade radius, then shine the Rydberg-exciting laser. The whole gate turns on one fact: only one atom can be excited at a time.
The target's fate is therefore conditional on the control — precisely what a
controlled gate needs. Tuned correctly this realises a controlled-phase (and hence a
The second platform goes the opposite way — not bigger and freer, but tiny and
factory-made. A quantum dot is a nanoscale pocket in a semiconductor,
shaped by tiny gate electrodes, that confines a single electron. The qubit is that
electron's spin: spin-up is
The appeal is the manufacturing base. A quantum dot is a whisker of the size of a transistor, and it is
built out of the same silicon and the same CMOS lithography that already prints
billions of transistors per chip. If qubits can ride that industrial process, scaling to millions could
follow the path the whole computing industry has walked for fifty years. Coherence is a bonus:
isotopically-purified silicon (stripped of the magnetic
A silicon spin qubit is roughly tens of nanometres across — millions of times smaller in area than a superconducting qubit. Read that two ways.
So the very size that makes spins scalable also makes them hard to connect, and packing millions of dots demands fabrication uniformity (each dot nearly identical) and a solution to the wiring/control problem — getting a control line to every qubit without drowning the chip in cables. That tension — glorious density, awkward coupling — is the central story of silicon spin qubits.
Mapping each platform onto the
The two platforms are almost mirror images of one another, and it is worth holding both pictures at once. On one side, atoms are trapped in beams of light and shuffled into any pattern you please — a square lattice, a ring, a bespoke graph — with each atom a flawless copy of the next and any atom reachable, via the blockade, across a whole radius (or physically shuttled next to its partner mid-circuit). Connectivity is a design choice, not a constraint. On the other side, qubits are etched into silicon by the same machines that make the chip in your phone, so scaling rides fifty years of industrial momentum — but each qubit is a speck that can only whisper to the dot immediately beside it. Freedom-and-scale-by-light versus density-and-manufacturing-by-silicon: two very different bets on how to get to a million qubits.
It is tempting to look for a single "winning" technology, but each platform simply moves the hard part somewhere new. Neutral atoms buy you scale and flexible geometry almost for free — but the entangling gate, the Rydberg blockade, is the delicate step: gate speed and fidelity are still maturing, and atoms occasionally get lost from their traps and must be reloaded. Silicon spins are wonderfully tiny and factory-friendly — but their interactions are very short-range, and packing millions of dots demands fabrication uniformity and a way to wire and control every one of them. Do not read "hundreds of atoms" or "rides CMOS" as "already solved." Every platform — atoms, spins, superconductors, ions — is a different trade-off, strong somewhere and paying for it elsewhere. As of today, no platform dominates.
Neutral atoms and silicon spins are two fast-rising ways to build a quantum computer, each acing the DiVincenzo checklist in its own way and each with a signature strength and a signature weakness.