Somewhere in a Texas Instruments lab, there is a sealed cavity the size of a grain of pepper with a whiff of foul-smelling gas trapped inside. The gas is carbonyl sulfide, the compound that gives rotten eggs some of their competition. It is not there by accident. When you hit that molecule with a millimeter-wave signal tuned to exactly 231 gigahertz, it starts to spin in a way that never varies, anywhere in the universe, forever. TI has been quietly patenting that spin.

For seventy years, an atomic clock has meant more or less one thing: a chamber of cesium atoms, a laser to interrogate them, and a rack of optics to keep the laser honest. Shrink all of that onto a chip and you get the chip-scale atomic clock, a genuine engineering marvel that still costs about as much as a used car engine and still needs a tiny laser. What TI has been filing, patent after patent, is a clock with no atoms and no laser at all. It runs on the same millimeter-wave circuitry that goes into 5G radios and automotive radar. And almost nobody outside the RF-chip world has noticed that the largest pile of US patents on this idea belongs not to a defense contractor or a quantum startup, but to the analog-chip company best known for graphing calculators.

The number

Since 2019, Texas Instruments has been granted 14 US patents that describe atomic clocks, more than any other company in the dataset. Honeywell, the traditional heavyweight in precision timing and navigation, has seven. But the raw count undersells how concentrated TI’s bet is. Ten of those patents describe a specific architecture the company calls a molecular clock: titles like “Molecular atomic clock with wave propagating rotational spectroscopy cell” and “Millimeter wave chip scale atomic clock.” On the narrower search for chip-scale atomic clocks specifically, TI leads with five grants since 2015 to Honeywell’s four, with the rest scattered one apiece across Boeing, Teledyne, Southwest Research Institute, and a handful of Korean labs.

Read the claims and the coherence is unmistakable. These are not fourteen unrelated filings that happen to share a keyword. One 2023 patent describes “a gas cell having a sealed interior” with “a dipolar gas inside,” probed by two electromagnetic waves sent through the cell in opposite directions, with a receiver measuring how much energy the gas absorbs. A 2021 patent describes “a hermetically sealed cavity” holding “a dipolar molecule that exhibits a quantum rotational state transition at a fixed frequency,” feeding clock-generation circuitry built from a detector, a multiplier, and a reference oscillator. A 2024 patent adds the control loop that keeps the whole thing locked. Strip out the word “clock” and every one of these describes the same machine: a sealed pocket of polar gas, a millimeter-wave source, and a feedback circuit that hunts for the exact frequency where the gas drinks the most energy. That frequency is the tick.

Why a spinning molecule beats a cesium atom

The reason this matters is cost, and the reason it might matter enormously is that the world is suddenly very short on trustworthy time.

Every GPS satellite carries an atomic clock. The entire positioning system is, underneath, a way of broadcasting extremely precise time; your phone computes location by comparing timestamps from satellites overhead. Jam or spoof that signal and a device is blind, unless it carries its own clock good enough to coast through the outage. Since 2022, GPS jamming has gone from a niche military concern to a daily fact of aviation over Ukraine, the eastern Mediterranean, and the Persian Gulf, where thousands of commercial flights a month now report degraded navigation. A clock that can hold accurate time on its own, with no signal from the sky, is the fallback. The market for chip-scale atomic clocks, valued at $1.2 billion in 2024 by Verified Market Research, is projected to reach $2.1 billion by 2033 on exactly this demand.

The incumbent design comes straight from a 2004 breakthrough at the National Institute of Standards and Technology, where John Kitching’s group built the first atomic clock from microfabricated parts, its guts the size of a grain of rice. Symmetricom turned it into the SA.45s, the first commercial chip-scale atomic clock, in 2011; the product now lives inside Microchip Technology. It works beautifully. It also has a laser, a cesium cell, and optics that must stay aligned, and its price has fallen from roughly $3,000 in 2011 to just under $1,000 today. Under $1,000 is cheap for a cruise missile and absurd for a phone.

This is where the adjacent possible sneaks in. The insight behind TI’s filings, worked out in the open by an MIT group the company was funding, is that you don’t need atoms at all. You need any quantum transition that is fixed by the laws of physics and that you can interrogate with electronics you already know how to mass-produce. A molecule’s rotation fits. “We searched for different physics all together,” MIT’s Ruonan Han told MIT News when his lab published the first fully electronic molecular clock in Nature Electronics in 2018. “We don’t probe the behavior of atoms; rather, we probe the behavior of molecules.” The molecule was carbonyl sulfide, chosen because it has a clean rotational line at 231 gigahertz, a frequency that CMOS millimeter-wave circuits can now generate directly. “The output of the system is linked to that known number,” Han explained. “You want to correlate a quantity that is useful to you with a quantity that is a physical constant, that doesn’t change.”

The MIT work was supported in part by a Texas Instruments fellowship. What was a physics demonstration in an academic paper shows up, a few years later, as a wall of TI patents describing the same sealed gas cell and the same millimeter-wave interrogation, engineered for a package you could solder to a board.

Who cares

Han said the quiet part in his own interview: “In the future, you don’t need to spend a big chunk of money getting atomic clocks in most equipment. Rather, you just have a little gas cell that you attach to the corner of a chip in a smartphone, and then the whole thing is running at atomic clock-grade accuracy.” A laser-free clock built from the same process node as a radar chip is the kind of thing that could eventually cost dollars instead of hundreds of dollars. At that price, atomic-grade holdover stops being a defense-procurement line item and starts being a checkbox feature in drones, base stations, data-center switches, and eventually handsets, all of which need precise time and all of which are increasingly assumed to operate through GPS outages.

None of this means TI is about to ship a molecular clock; a patent portfolio is a description of what a company has built and protected, not a product roadmap. But it does reframe who is positioned. The reflexive assumption is that resilient timing is a game for Microchip, Honeywell, and the national labs. The patent record says the company assembling the deepest position in the post-laser architecture is a mixed-signal chipmaker that already owns the millimeter-wave manufacturing base the whole approach depends on. When timing finally gets cheap enough to put everywhere, the low-cost version may not come from the atomic-clock industry at all. It may come from the radar aisle.

Method note

Counts come from a corpus of 9.3 million US utility patent grants sourced from USPTO bulk grant data, searched by full text for atomic-clock and molecular-clock terminology and grouped by assignee, with variant spellings and subsidiaries combined. The 14-patent figure covers grants issued from January 2019 through mid-2026; the chip-scale comparison covers 2015 onward. Patent claims and abstracts quoted here were read directly from the grant texts. The origin science is drawn from the 2018 Nature Electronics paper by Cheng Wang, Ruonan Han and colleagues at MIT, and from MIT News. Market figures are from Verified Market Research; the GPS-jamming context and the NIST/Symmetricom history are from NIST, IEEE Spectrum, and US Army C5ISR reporting. A patent portfolio documents what has been filed and protected; it is not evidence of a shipping product or a stated commercial plan.