Home » Quantum Computing » Osaka’s Photonic Chip Couples Ten Atom Sites to Waveguides

Osaka’s Photonic Chip Couples Ten Atom Sites to Waveguides

Osaka researchers built a glass chip that reads photons from ten atoms in parallel. It is groundwork for linking quantum processors, not yet a link between them.

Abstract blue quantum photonic interconnect network of glowing light lines and nodes representing linked quantum processors

Useful quantum computers may need far more qubits than any single chip can hold. Osaka University researchers built a piece of hardware that could help link smaller machines. Their quantum photonic interconnect is a glass chip that collects photons from ten atoms in a neutral-atom array through ten parallel waveguide channels. It replaces bulky fiber-optic arrays. It does not yet link ten separate processors. Tech Times reported the result on September 1, 2026.

Who built it, and did they link ten quantum computers?

No, they didn’t. Professor Takashi Yamamoto led the team at Osaka’s Center for Quantum Information and Quantum Biology, working with Japan’s National Institute of Information and Communications Technology and Hamamatsu Photonics. The chip collected photons from ten atom sites in one neutral-atom array. The team says the approach could help connect quantum computers later. The work was published in Optica.

Why a glass chip instead of fiber?

Spacing. Fiber arrays space their channels about 127 micrometers apart or more, which is too wide to match how tightly researchers pack atoms in a neutral-atom array. The atoms here sit 7.5 micrometers apart, and the optics map that spacing onto the chip. Yamamoto’s chip is a glass device with a 32-channel waveguide array, spaced 25 micrometers apart, and ten channels were used in the test. Glass was chosen because silicon absorbs light at the 780-nanometer wavelength used here, per Tech Times.

Earlier fiber-based setups handled only a few channels. Tech Times says the prior ceiling was four or five.

What did the chip actually measure?

It collected 780-nanometer photons from ten atom sites at once. Each photon’s polarization was correlated with the internal state of the atom that emitted it. The team reports a visibility of 0.87. That number measures how strongly an atom’s state matches its photon’s polarization, and Tech Times notes that a value above 0.5 rules out classical correlations. Detectors caught photons with 75% to 94% efficiency per channel, and crosstalk between channels was negligible.

Osaka’s setup reads the photons with a 12-channel superconducting nanostrip detector system built by NICT and Hamamatsu. Detector work like the one NIST recently reported with much wider nanowire detectors is a related effort, not the same device.

Why It Matters to You

Nobody can use this today. It is lab hardware, groundwork for linking quantum machines, and no application built on it exists yet. If it scales, the people who would feel it first are researchers building larger quantum computers, and anyone who hopes those machines will one day help with problems like rerouting ships around storms. Picture Priya, an illustrative example and not a real person, a logistics analyst with that hope. It is a hope, not a result.

Why do scaling and linking matter?

Fault-tolerant quantum computers may need more than one million qubits. Neutral-atom arrays are expected to reach about 10,000 atoms, according to Nanowerk’s summary of the Osaka release. So researchers plan to wire many smaller processors together. That only works if a photonic interconnect can move quantum information between them without losing it to noise.

Other teams attack the qubit shortage differently. Researchers at Singapore’s A*STAR described a method to pack more of a calculation onto fewer qubits, an early preprint tested on IBM hardware. Osaka’s approach aims to connect many chips instead. Both paths try to get around the limit of a single processor.

What’s the catch?

Several things. The demonstrated link is between each atom and its own photon, and the sources report no entanglement between two separate processors. The chip handled ten atom sites in one array, with ten of 32 waveguide channels in use. Coupling efficiency is comparable to earlier fiber experiments, per Tech Times, so the chip fixed the density problem but not the rate problem. Entanglement stays probabilistic and slow. The “first demonstration of a multiplexed optical interface” and “most channels ever” claims come from the team and Tech Times, with no independent expert review found. And the paper appeared first as an arXiv preprint in December 2025 and then in Optica, but the sources here are the preprint abstract and press coverage, not the full journal paper.

What happens next?

The team published its results in Optica on September 1, 2026. Its next target is about 100 channels, a goal and not a result. Tech Times says the main engineering hurdle is coupling efficiency, the share of emitted photons that reach a detector. Low coupling means many attempts per entangled pair, which limits networking speed. Yamamoto said the team will increase the number of channels and work toward connecting neutral-atom quantum computers. Chips like this are plumbing, not a finished network.

Sources

Related Reading