A 37-person chip company in Sunnyvale is quietly trying to retire the GPS signal your phone has used since 1995.
Somewhere over the Pacific, a Galileo satellite is transmitting a signal that almost nothing on Earth bothers to listen to. The signal is called L5. It is wider, cleaner and harder to fool than the L1 signal your phone uses, and it has been broadcasting since 2009. For sixteen years, the chip industry has treated it as an interesting accessory.
On a quiet block in Sunnyvale, in a building that used to belong to someone else, a small team is treating it as the whole point.
The chip on the bench is a few millimeters wide. The roadmap on the wall does not include the L1 band at all.
oneNav was incorporated in late 2019 by two people who had each built large parts of the GPS chip industry. Steve Poizner co-founded SnapTrack in the 1990s, the company that put GPS in a cell phone for the first time and was acquired by Qualcomm in 2000 for a billion dollars. Paul McBurney co-founded eRide, ran GNSS architecture at Apple, and holds more than forty location patents.
They started oneNav around a single observation. The L1 band is congested, easy to jam, and easy to spoof. The L5 band is none of those things. The conventional move - the move the entire industry made - was to add L5 as a second channel on top of L1. oneNav decided to take L1 off the chip.
The architecture is called L5-direct. It acquires and tracks the L5 signal without using L1 as a starting point. That single decision halves the RF footprint, halves the power, and changes the math on jamming resilience by roughly an order of magnitude.
The company sells the design two ways. The first is the pureL5 digital IP core, licensed to chipmakers who integrate it into their own SoCs. On a 3-nanometer process it occupies about 0.28 mm² and draws around 4.7 milliwatts in tracking mode. The second is the L5-direct ASIC, a finished chip fabricated on GlobalFoundries' 22FDX FD-SOI platform, brought up for the first time in May 2025.
The customers fall into two camps. Commercial buyers want a smaller, cheaper GNSS block for phones, wearables and IoT modules. Defense buyers want a receiver that does not fall over when somebody points a jammer at it. In-Q-Tel - the investment arm associated with the U.S. intelligence community - sits on the cap table, and a Direct-to-Phase II SBIR award from the Department of Defense arrived in May 2025.
oneNav is not chasing scale by hiring fast. It has roughly 37 people. Most of them have built GNSS silicon before, at Qualcomm, Apple, Intel, SiRF, Trimble or SnapTrack.
The civilian GPS signal you've been using since the 1990s rides on the L1 band. The newer signal - L5 - was designed once the engineers had three decades of hindsight. oneNav's argument is that holding on to both is the timid choice.
Co-founded SnapTrack, the company that integrated GPS into mobile phones; sold to Qualcomm for $1 billion in 2000. Served as a White House Fellow on the National Security Council, and as California Insurance Commissioner. oneNav is his fourth Silicon Valley startup.
Previously co-founded eRide (acquired by Furuno) and ran GNSS architecture at Apple. Career in receiver design across Stanford Telecom, Trimble, eRide and Apple. Holds 40+ location-related patents. PhD in Electrical Engineering, Iowa State.
A licensable single-frequency L5 GNSS receiver core. Roughly 0.28 mm² in a 3nm process. Customer-evaluation systems ship to chipmakers integrating it into their own SoCs.
The first finished L5-direct receiver chip. Fabricated on GlobalFoundries' 22FDX FD-SOI process. Brought up in May 2025.
Positioning, Navigation and Timing module aimed at military, government and critical-infrastructure use in contested or jammed environments.
Real-time satellite monitoring for telecom operators and non-terrestrial network deployments where signal integrity is the product.
A precision augmentation product launched alongside the Series C round, layered onto the L5-direct receivers for mass-market accuracy.
An on-device model that distinguishes direct satellite signals from urban reflections - the failure mode that quietly ruins GPS in dense cities.
Steve Poizner and Paul McBurney incorporate oneNav around an L5-only architecture thesis.
The technical case for skipping L1 is published.
Norwest Venture Partners and GSR Ventures join. Total raised reaches $33M. World's first pure L5 mobile GNSS receiver is unveiled.
GV, GSR, Norwest and In-Q-Tel return. pARC augmentation product launches.
Fabricated on GlobalFoundries 22FDX. Simultaneously, the U.S. DoD awards a Direct-to-Phase II SBIR for Direct-to-L5 technology.
Most people meet GNSS at its worst moments. A blue dot drifting in a city block. A rideshare driver pinned to the wrong street. A delivery van that thinks it's on the freeway above instead of the surface road below. These are not edge cases. They are multipath - reflected signals bouncing off glass, and they are the silent failure mode of legacy GPS.
oneNav's pitch is that L5, used directly, behaves better in those moments. The signal is wider and cleaner. The chip can use AI to distinguish a real satellite signal from its reflection. The receiver is smaller and uses less battery, which matters when the device is a wearable.
The other side of the pitch is harder to test in a parking garage. A consumer-grade GPS receiver is also the receiver that flies in drones, sits inside autonomous trucks, and times every cell tower and substation in the country. When that signal is jammed or spoofed - which now happens regularly in conflict zones and increasingly in commercial airspace - the failure modes get expensive.
An L5-direct receiver is much harder to jam. That is roughly why In-Q-Tel is on the cap table and the SBIR award was Direct-to-Phase II.
Alphabet's venture arm, lead investor on the 2021 round.
Led the $17M Series C in late 2023.
The investor that brings U.S. defense and intelligence buyers to the table.
Multi-round backer.
Multi-round backer.
Process partner: 22FDX FD-SOI for the L5-direct ASIC.
The satellite over the Pacific is still transmitting L5. It is still being ignored by almost every chip on Earth.
But the chip on the Sunnyvale bench has tape-out behind it, a foundry partner, an SBIR award, and a cap table that includes an Alphabet fund and a CIA-affiliated fund on the same line. The roadmap on the wall still does not include L1.
If oneNav is right, the next time your phone draws a blue dot in a parking garage, the dot will be where you actually are. If oneNav is also right about the second thing - the harder thing, the thing the SBIR award is for - the next time someone tries to point a jammer at a drone, the drone will not care.