The Honest State of UWB: What the Market Reports Skip
If you are jumping into Ultra-Wideband (UWB), think carefully about your choices.
There is a lot of change coming and a lot of constraints to work around, and the wrong architecture or an unstable supply chain can paint you into a corner you spend a whole product cycle climbing out of.
The market reports sprayed across the internet will not warn you about any of this. Instead, they tell you the technology is growing fast: close to 450 million UWB chips were shipped in 2024, an increase of 21% year over year. They confirm centimeter-level accuracy. They list the same handful of reference deployments. But none of that helps you decide.
Earlier in our series on UWB (UWB vs BLE vs WI-FI), we covered how it outperforms BLE today and what 802.15.4ab will bring next. The harder question to answer is this one. You need to know whether your chip supply will hold for the next two years. Whether your product will need three different hardware variants to ship globally. Whether your operations team can maintain dedicated UWB infrastructure once the pilot ends.
Those are the questions the market reports skip. This article covers them.
Where UWB is Actually Deployed (Not Just Announced)
Let’s start with what is working. Consumer tags like Apple AirTag and Samsung SmartTag use UWB for precision finding at close range, backed by BLE for broader discovery. Also, BMW, Genesis, and Mercedes ship UWB-based digital car keys that unlock as you walk toward the vehicle. Elsewhere, Volkswagen Slovakia runs a factory-floor tracking system with sub-second position updates and 99.8% reliability. All three run at production scale today.
Now look at the pattern. Consumer tags, automotive keys, and select factory RTLS. Three categories. That is the honest list of where UWB is running in production today, despite a decade of industry effort.
Think of UWB chips today the way you would think of Bluetooth chips in 2005. Almost every laptop shipped with one. Almost no one used it. The hardware was in place years before the software, the standards, and the user habits caught up. UWB is at that stage now. Roughly 60% of UWB chips end up inside smartphones, where they mostly sit dormant because the apps and services that would use them barely exist outside of Apple’s and Samsung’s ecosystem.
Also, chips in phones are different from deployed systems. If you strip out consumer electronics, the number of “operational UWB installations” in industry and enterprise shrinks to a small set (a handful of automotive OEMs, a thin slice of factory RTLS pilots, and the early access-control programs).
UWB is working, but only in a few specific use cases so far. Adoption is concentrated, and to move from those three working categories to the dozens of use cases in the market reports has real obstacles. Like chip supply, regulation, and the gap between pilots and full deployments.
Only Two Viable Suppliers
Here is a fact that rarely appears in UWB thought leadership: if you are building a production UWB product today, your real chip options come down to NXP and Qorvo.
Let’s put that into perspective. Apple designs its own UWB silicon (the U1 and U2 chips) and does not sell it. Samsung’s Exynos Connect U100, launched in 2023, is marketed externally to automotive and IoT customers, but few companies outside Samsung have publicly adopted it.
That leaves NXP and Qorvo splitting the rest of the production market between them. That is a problem. Both supply the same three markets — automotive, industrial, and consumer — and lead in different corners of each. NXP leads in automotive secure entry through its SR1xx and Trimension families, the silicon line BMW uses for Digital Key Plus.
Qorvo’s DW3xxx series covers a wider mix: industrial RTLS, IoT tags, and CCC Digital Key 3.0-compliant automotive deployments. In March 2025, Qorvo announced its first fully integrated UWB system-on-chip, putting the radio and processor on one chip instead of two.
Building a product on a two-supplier market is like running a restaurant with two flour mills in the region. On a normal week, you have a backup. The day there’s an issue with one supplier and the other raises prices because they can. Then your menu and your margins are at the mercy of whoever still has stock. Two suppliers is workable, but the constraint shows up on a two-year product development cycle. Pricing leverage sits with the vendor, not with you.
If one supplier has a production issue or a fab allocation conflict, your second-source options are limited. And because each vendor’s SDK and ranging protocol stack are proprietary, switching between them mid-project means you may have to redesign your entire stack.
Chinese chipset vendors are entering the space. But “entering” and “production-qualified with a stable SDK and field-proven ranging performance” are separated by years, not months.
For teams making chip-down decisions right now, the honest answer is: you are choosing between NXP and Qorvo. Compare that to BLE, where a team can choose from at least seven established silicon vendors: Nordic, Silicon Labs, TI, Espressif, Infineon, NXP, and STMicroelectronics. Wi-Fi offers a similar field. UWB gives you two. That is a level of vendor dependency most embedded radio decisions never carry.
The Regulatory Tax Nobody Calculates
Every UWB article mentions “regulatory challenges” as a bullet point. None of them spell out exactly how and what that costs your product in practice. For teams building a global product on a startup timeline, that “bullet point” costs you on two fronts: six figures to build and certify the regional variants, and six months on your schedule. And it is rarely in the project plan until it is too late to absorb.
In August 2025, China’s MIIT issued interim provisions that ban UWB operation on the lower frequency channels and mandate channel 9 (centered at approximately 8.0 GHz) for commercial devices. Europe, through Implementing Decision 2024/1467, increased allowable indoor UWB transmission power by 10 dB and opened certain outdoor applications. The United States has been discussing similar updates but has made no formal decision.
Now translate those three regulatory positions into a product roadmap.
If your device targets all three markets, as most do, you are potentially looking at different channel configurations per region. Different channels can mean different antenna tuning, different RF front-end matching, and different certification test campaigns. What that means is: a tag running on channel 5 in the EU and channel 9 in China cannot use the same antenna design. You either add a tunable matching network that handles both frequencies, or you build two versions of the product
Each variant also needs its own certification. UWB certification is not just FCC or CE radio testing. It includes IEEE 802.15.4z compliance and, increasingly, FiRa consortium interoperability testing. Budget two to four months per region for the full certification path. Multiply that by the number of hardware variants. Yikes!
Why Pilots Work and Deployments Stall
A UWB pilot is a wind tunnel test. A production deployment is the open road.
The most common UWB success story reads like this: an x company runs a pilot in a controlled area. Either a warehouse zone, a hospital wing,or a section of a factory floor. The pilot works. Centimeter accuracy, fast updates, impressive demo. The team presents results, secures budget for full deployment, and then the project stalls.
The reasons are operational, not the radio physics most people would point to first. The physics did not change between pilot and production. What changed is scale, environment, and time.
Calibration does not survive contact with production: UWB positioning depends on precise time-of-flight measurements. A one-nanosecond error in the ranging timestamps translates to roughly 30 centimeters of position error. Every UWB tag ships with an antenna delay value that must be calibrated per unit to maintain accuracy. For instance, in a pilot with 20 tags, you calibrate each one on the bench and move on. In a production deployment with 2,000 tags, that calibration step becomes a factory process that needs fixtures, test procedures, and pass/fail criteria. Teams that skip this step, or assume factory defaults are close enough, can kiss their centimeter level accuracy goodbye.
Environments change: Anchors are calibrated to a specific physical layout. When a warehouse rearranges shelving, installs new metal racking, or adds a mezzanine level, the multipath profile of the radio environment shifts. UWB’s short wavelength (roughly 3.76 cm at 8 GHz) makes it particularly sensitive to reflections off metal surfaces. A pilot environment is static. A production environment is not.
Anchor infrastructure needs ongoing maintenance: UWB anchors are not Wi-Fi access points that building IT already manages. They are dedicated infrastructure that needs power, mounting, clock synchronization, and periodic health monitoring. The operational cost of maintaining an anchor network over years is routinely underestimated. When the facilities team that owns the building does not own the anchor network, maintenance gaps appear. And when maintenance gaps appear, position accuracy drifts silently until someone files a support ticket.
For teams planning a UWB rollout, you need to budget for the operational team, the calibration process, and the anchor maintenance program before you budget for the radios. That is where production deployments succeed or stall.
Two Developments to Watch, Not to Plan Around
IEEE 802.15.4ab: IEEE 802.15.4ab is the next-generation UWB standard currently in development. An earlier piece in this series broke down what 802.15.4ab adds to the radio’s job description: standardized radar sensing, up to 30x range extension, and 18x higher data rates. Also, the standard is backward compatible with 802.15.4z, so existing infrastructure and chips will continue to operate. The open question for product teams is timing: when will silicon implementing 802.15.4ab be available in production quantities? If you are building a product that ships in 2027, the advice is: design for 802.15.4z today and leave room in your architecture for a radio swap later.
Chip ecosystem expansion: is the other trend. SPARK Microsystems, STMicroelectronics, and several Chinese semiconductor companies are entering the UWB chip market with new architectures. More suppliers means more pricing competition, more second-source options, and potentially lower barriers for smaller product teams. But new entrants need time to build stable SDKs, field-test their ranging performance across real indoor environments (not just anechoic chambers), and earn the design wins that prove their silicon in production. A chip that demos well at CES is not the same as a chip your operations team can procure at volume pricing with a 52-week delivery commitment. Do not plan your next product around a chip that has not shipped in volume yet.
The Technology Works, The Ecosystem Is Catching Up
Don’t get me wrong, UWB is not overhyped by any means. The radio physics delivers what it promises: centimeter-class ranging, secure distance measurement, and performance that no other wireless technology matches at close range.
What is lagging behind the physics is everything else.
The chip supply is concentrated. Regulation is fractured across regions and still moving. The operational reality of maintaining UWB infrastructure at production scale is harder than the pilot made it look. And the content written about UWB keeps recycling the same accuracy spec and market projection without addressing any of it.
If your product needs centimeter positioning, UWB is the right radio. But “right radio” is only the first decision. The second, third, and fourth decisions involve how you will source your chips, how many hardware variants your regulatory strategy requires, and whether your organization is prepared to own and maintain dedicated anchor infrastructure for years.
Those decisions are where embedUR works with product teams. We map the engineering reality of UWB adoption to your specific product timeline, factory process, and target markets — before the surprises show up in month six..
Which costs more for your team: a 30-minute architecture review now, or six months of unplanned regulatory rework later?
