Sixty-two kilometers of ordinary internet cable, the kind exposed to wind and Maryland's summer heat, carried a pair of linked light particles between two labs, and the link between them held for a full day. Picture two magic coins. Flip one in Maryland, and the coin in California lands the opposite way at the same instant, every time, with no call placed and no signal sent between them. That is entanglement: a bond between two particles that acts as though distance does not exist.
Researchers built exactly that bond, then tried to break it.
A Photon Stayed Home. Its Twin Went to College Park.
The National Institute of Standards and Technology, working with the University of Maryland and quantum networking company Qunnect, generated pairs of linked photons in a lab in Gaithersburg. One photon from each pair stayed put. Its twin traveled 62 kilometers through commercial fiber, the same kind of cable that carries phone calls and video streams, to a second lab at the University of Maryland in College Park. NIST published the results with its collaborators in the Journal of Optical Communications and Networking.
The Fiber Was Never Built for This
The cable strung above the street in this test is ordinary telecom infrastructure, hung on poles, not buried in a shielded line built for quantum signals. That distinction is the whole story for a network architect. Building dedicated fiber for a new technology costs money nobody spends until the use case is proven, and quantum networking has been stuck waiting on that proof for years. NIST's test shifts the burden onto anyone who wants to build a new quantum network from scratch instead of leasing the cable already hanging overhead.
A Wobbling Cable Nearly Ruined the Signal
Phone calls and video streams travel as changes in brightness. A cable swaying in the wind or heating up in the sun barely touches that. Entangled photons carry their link differently, in something called polarization, essentially the angle the light wave leans at as it moves. That angle shifts every time the cable expands, contracts, or sways, and an exposed cable never stops doing all three.
NIST physicist Oliver Slattery called the aerial cable "about as bad a connection as you can possibly have" (NIST, 2026). His team ran the test on it anyway. Over 24 straight hours, 92.8 percent of the entangled pairs arrived still correctly linked (NIST, 2026).
A Second Beam of Light Fixed the First, in Real Time
Fixing that shifting angle meant sending a second beam of light down the same cable to measure and undo the damage. NIST's team used Qunnect's QU-APC, an automated polarization compensator the Brooklyn company has sold commercially since 2023 (Qunnect, 2023). The device sends a plain reference beam through the fiber alongside the entangled photons, measures how much the cable has bent that beam, and twists the entangled photons back the opposite way before anyone reads them. It runs at room temperature, which matters, because rival approaches that need deep-freeze cooling are not something an enterprise network team could plug in and operate.
Qunnect has advertised uptime above 95 percent for QU-APC in past demonstrations, and a 2024 test on the company's own GothamQ network in New York hit 99.84 percent over 15 days (Qunnect, 2024). NIST's independent test landed lower, at 92.8 percent, on fiber the company does not control, in conditions no press release gets to pick. Remember that gap the next time a vendor's uptime number shows up in a slide deck.
The Distance Record Still Belongs to Buried Cable
A European team sent entangled photons 248 kilometers through underground fiber in 2022, four times farther than NIST covered here. Underground cable sits at a stable temperature and does not move, so distance was never the hard part of that experiment. Exposure is the hard part. NIST's route sat above ground for its entire length, taking the heat and wind a real city network would take, and still delivered a linked signal 92.8 percent of the time (NIST, 2026).
The lab is not the finding. The commute is.
The finding does not solve the last-mile problem in quantum networking. It reframes who has to solve it.
Qunnect Has Already Sold This Story to Investors
Airbus Ventures led a $10 million funding round for Qunnect in June 2025, with Cisco Investments and Quantonation joining in (Qunnect, 2025). The company runs a 34-kilometer entanglement testbed across Brooklyn and Queens called GothamQ, and it has deployed a commercial network in New Mexico with Roadrunner Venture Studios. A seven-year-old hardware startup with defense-adjacent backers and a live network already running is a different bet than a university spinout still chasing its first contract.
The Same Landlord Logic Is Already Playing Out in Compute
Data center operators worked out this arithmetic before Qunnect did. In India, CoreWeave and five other Nvidia-backed neoclouds are leasing capacity from companies that already own the land and the power contracts, Sify, Yotta, CapitaLand, CtrlS, and Tata Communications among them, instead of building new sites (shashi.co, July 2026). Applied Digital ran the same play in North Dakota, delivering finished capacity to CoreWeave on schedule rather than CoreWeave building its own facility. Telecom carriers holding fiber they already depreciated sit in the identical spot: landlords to a new kind of tenant, not builders of a new network.
NIST validated the physics. It also validated the same kind of asset infrastructure investors have been paying up for in AI compute all year.
If a fragile quantum signal survives on unshielded fiber your carrier already owns, the case for building a brand-new quantum network shrinks to a leasing negotiation, the same shift already underway in GPU capacity. Ask your network provider whether they have tested their aerial routes, the cable strung on poles above ground, for this kind of signal. Then ask Qunnect or a comparable vendor whether their hardware has been validated on the specific routes your data centers already lease, not only on a network the vendor controls.
Terms, in Plain Language
Entangled particles. Two particles linked so that measuring one instantly tells you the state of the other, no matter how far apart they are.
Photon. A single particle of light.
Polarization. The angle a light wave leans at as it travels. Entangled photons carry their link in this angle, which is why a twisting cable can break the connection.
Aerial fiber. Cable strung on poles above ground, exposed to weather, rather than buried underground where conditions stay stable.
QU-APC. Qunnect's device for this test. It measures how much a cable has distorted a light signal and corrects for it in real time.
Uptime. The share of the time a system works as intended. Here, the percentage of entangled photon pairs that arrived still correctly linked.
Neocloud. A cloud provider built around one product, renting out computing chips by the hour, instead of the wider menu a company like AWS or Google Cloud sells.
Shi, Yicheng, et al. "Entanglement Distribution Over a Polarization-Stabilized Aerial Fiber." Journal of Optical Communications and Networking, 15 July 2026, opg.optica.org.
Qunnect. "Quantum Networking Pioneer Qunnect Raises $10 Million in Oversubscribed Series A Extension." Qunnect, 24 June 2025, www.qunnect.inc.
Qunnect. "Qunnect Debuts QU-APC Instrument with Interactive Demo at the 2023 Optical Fiber Communication Conference and Exhibition." PR Newswire, 13 Mar. 2023, www.prnewswire.com.
Qunnect. "Qunnect Achieves Record-Breaking Performance for Distributing Polarization Qubits on GothamQ Network in NYC." PR Newswire, 15 Apr. 2024, www.prnewswire.com.
Bellamkonda, Shashi. "CoreWeave and Five Other Nvidia-Backed Neoclouds Seek to Lease India Data Center Capacity." shashi.co, 29 July 2026.
Bellamkonda, Shashi. "The $23 Billion Backlog Nobody Can Name." shashi.co, 24 Apr. 2026.
