There is a pane of glass in a Tokyo office tower that lies about being a window. From the inside it looks like any other sheet of float glass. But printed across its surface, in a grid too fine for the eye to resolve, sit thousands of metal cells tuned to the rhythm of a 28-gigahertz radio wave. Slide the glass a fraction of a millimeter and the pane changes its mind about the signal passing through it: let it all in, bounce some of it sideways down a corridor, or split the difference. NTT DoCoMo and the glassmaker AGC demonstrated that window in 2021, and they called it the world’s first transparent dynamic metasurface. DoCoMo’s pitch, reported at the time by Light Reading, was blunt: stop building cell towers. Turn the windows, the streetlights, and the traffic signals you already own into radio infrastructure.

That window is the friendly face of a technology with an unfriendly acronym: the reconfigurable intelligent surface, or RIS. Strip the jargon and it is a wall that reprograms how radio waves bounce off it. A normal wall scatters a signal wherever physics dictates. A RIS is a flat array of tiny tunable elements, each able to nudge the phase of the wave that hits it, so the wall as a whole can aim a reflection at a specific phone in a specific dead zone. No amplifier, no radio chain, almost no power. Just a surface that has been taught to steer light it cannot see.

And in the United States patent record, that surface is having a breakout year. Across 9.3 million US utility grants sourced from the USPTO’s bulk feed, grants describing reconfigurable intelligent surfaces in a wireless context went from a single grant in 2019 to 69 in 2025. The 2026 count is already past last year’s total with half the year still to run. Two hundred eighty-one of these grants now exist where almost none did five years ago.

Here is the part nobody is reporting. The company winning that race is not the one you would guess.

The surprise on the assignee line

Ask who owns RIS and a wireless engineer will name European and Chinese academics, or Huawei, which has poured research into the idea. The citation record backs that instinct. The paper that lit the fuse, a 2018 study on using these surfaces for energy efficiency by Chongwen Huang, Alessio Zappone, Mérouane Debbah and Chau Yuen, has been cited more than 3,700 times. Marco Di Renzo’s framing of “smart radio environments” became the field’s catchphrase. The deeper root runs to Tie Jun Cui’s group at Southeast University in China, whose 2014 work on “coding” and “programmable” metasurfaces first proposed treating a sheet of metamaterial like a screen you could rewrite.

The US patent grants tell a different story about who is converting that science into property. Qualcomm holds 50 of them. Huawei has 6. Nokia has 4. Apple has 4. Qualcomm, in other words, owns more granted US patents on radio-steering surfaces than Huawei, Nokia and Apple combined, by more than three to one. Most of its haul landed in a two-year burst: 13 grants in 2023, 20 in 2024. The loudest voices in the literature are not the ones filing the deeds.

This is a familiar Qualcomm move. The company built its empire by owning the patents under a standard before the standard existed, then collecting a toll on every device that used it. CDMA made it rich. 5G keeps it rich. The 50-to-6 gap on RIS looks like the same playbook, run early, on a technology most of the press has not noticed.

Read the claims and the strategy changes

Count patents and you learn who is filing. Read them and you learn what they think the technology is for. This is where the Qualcomm portfolio stops looking like a coverage play and starts looking like something stranger.

Of Qualcomm’s 50 RIS grants, 18 are not about signal coverage at all. They are about location and sensing. One grant covers using a reconfigurable surface to calibrate the timing errors that corrupt a phone’s position fix. Another, titled “Peer-to-peer sensing assisted by reconfigurable intelligent surfaces,” describes configuring a surface to reflect signals between two devices so they can detect a physical object sitting between them. The wall is no longer relaying a call. It is a radar.

Then there is the one that reads like a spy novel. In a 2025 grant on “cooperative directional security,” a network picks a geographic zone it wants to protect, recruits one or more of these surfaces, and steers a deliberate wash of noise into that zone, while the real transmission goes out in a different direction entirely. An eavesdropper standing in the wrong place hears static. The intended receiver, somewhere else, hears the message clean. The same hardware that fills a dead spot can also draw an invisible cone of silence across a room.

The unifying thread is integrated sensing and communication, the 6G buzz-phrase for networks that locate and perceive the world while they carry data. A reflecting surface with a known fixed position is, geometrically, a free anchor point: bounce a signal off it and the extra path gives a receiver one more line to triangulate from. Qualcomm is not patenting better cell coverage. It is patenting the right to turn the built environment into a positioning and sensing grid, which is exactly the layer of 6G where a chipmaker that already lives inside every phone’s modem would most want to own the tollbooth.

The catch that makes this a bet, not a sure thing

Now the teeth. For all the filings, RIS does not have a standard to live in. The 3GPP, the body that decides what counts as 5G and 6G, considered making reconfigurable surfaces a formal study item for Release 18, then again for Release 19, and declined both times. It was left out of 5G-Advanced. The humbler cousin, a network-controlled repeater that simply amplifies and redirects a beam, got standardized instead. The patents are piling up ahead of any agreement that the technology belongs in a network at all.

There are reasons for the hesitation, and Emil Björnson has spent years naming them. In a 2020 paper with Özgecan Özdoğan and Erik Larsson, pointedly titled “Reconfigurable Intelligent Surfaces: Three Myths and Two Critical Questions,” he argued that the physics is harsher than the hype. A passive reflector pays a brutal price in path loss, because the signal weakens on the way in and again on the way out, and a large surface may need thousands of elements just to beat a cheap relay. The honest question, he wrote, is not whether these surfaces are clever but whether they ever earn their place against simpler hardware.

That is precisely what makes the patent landscape worth watching rather than dismissing. The science is genuine, the deployments are real enough that a Japanese carrier put one in a window, and yet the standards bodies keep flinching. Display makers are circling the same space from the other side. Japan Display, a company built on liquid-crystal panels, holds a granted US patent on an intelligent reflecting surface, and AGC’s glass version comes straight out of the microfabrication line that makes screens. The trick that aims a pixel and the trick that aims a radio wave are closer than either industry assumed.

If RIS does eventually land in a 6G release, whoever holds the foundational positioning patents will be standing at the gate when the standard arrives. Qualcomm has been to this gate before. The bet it is quietly placing, 50 grants deep, is that the radio mirrors are coming, and that the valuable part will not be the coverage. It will be knowing exactly where you are standing when the wave bounces back.

Method note. Counts come from a full-text search of 9.3 million US utility patent grants, sourced from USPTO bulk grant XML, for the phrases “reconfigurable intelligent surface” and “intelligent reflecting surface” in a wireless or communications context. The yearly series and the 281-grant total reflect granted patents only; the 2026 figure is a partial year. Assignee counts combine variant spellings and subsidiary filings for each company. The keyword filter will miss patents that describe the same hardware under other names, such as “smart radio environment” or “metasurface relay,” and a handful of unrelated optics patents share the “intelligent reflecting surface” phrase without affecting the leaders. Citation counts and publication years for the academic papers are from OpenAlex, which indexes roughly 357 million scholarly works. Deployment, standardization and skeptic details are drawn from NTT and Light Reading reporting on the DoCoMo-AGC trial, 3GPP release records, and the cited IEEE Communications Magazine paper by Björnson and colleagues.