For most of its history, the Bluetooth chip business has been an audio business. Earbuds, headphones and speakers bought the bulk of the silicon. New research suggests that balance is starting to tilt toward connected devices that send data rather than sound.
Japanese research house Techno Systems Research (TSR) estimates that Bluetooth IC shipments reached 8.3 billion units in 2025 and will climb to about 10.3 billion by 2032, based on its 2026 Wireless Connectivity Market Analysis. That is steady but unspectacular growth of roughly 3.2% a year. The more telling number is the mix: audio SoCs took 61% of shipments in 2025, and TSR expects the split between audio and data applications to move toward roughly even by 2031 to 2032.
What drives the data side is IoT: medical wearables, asset trackers, smart home products and digital car keys. And the chips going into those products are changing too, picking up precise ranging and on-device machine learning that turn a simple radio into something closer to a small edge computer.
Audio Is Maturing, Data Is Where The Growth Is
Audio has carried Bluetooth volumes for a decade, but TSR sees that segment slowing, with weakness at the low end of the smartphone accessory market weighing on shipments. True wireless earbuds are now a mature product category in most markets, and the upgrade cycle has lengthened.
Data applications are the opposite story. TSR describes connected health devices, tracking tags, smart home gear and automotive digital keys as the main growth engines for Bluetooth silicon over the rest of the decade. None of these product lines ships volumes on the scale of earbuds individually, but together they are broad, varied and still early in adoption.
Chips And Devices Are Counted Differently
Readers comparing forecasts will notice that the figures do not line up neatly. The Bluetooth Special Interest Group (SIG), which publishes yearly market data on the technology, projects Bluetooth device shipments of 6 billion in 2026, rising to 8.1 billion in 2030. ABI's own connectivity forecast from July put 2025 Bluetooth device shipments at 5.4 billion.
TSR is counting integrated circuits, not finished products. A pair of true wireless earbuds, for example, can contain more than one Bluetooth chip, so chip volumes naturally run ahead of device volumes. Both views point the same way: continued growth, with the fastest gains outside audio.
The SIG's category data shows where some of that momentum lies. It expects electronic shelf labels using Bluetooth to grow from 46.5 million units in 2026 to 175.7 million in 2030, a sign of how retail infrastructure is becoming a sizeable market for low-power radios.
Who Wins The IoT Side Of Bluetooth
The competitive picture in the IoT-focused part of the market looks very different from audio. In 2025 shipments of Bluetooth Low Energy single-mode SoCs, the chips that power most sensors, trackers and wearables, TSR puts Nordic Semiconductor at 21% of units and 36% of revenue. That gap between volume and value shows Nordic selling at higher average prices than most rivals.
Chinese suppliers, led by names such as Telink, PHYPlus, OnMicro and YiChip, took about 29% of shipments in the same category but only 9% of revenue. Their strength is cost, which suits high-volume, price-sensitive products. Other established players in the space include Silicon Labs, NXP, Texas Instruments, Realtek, EM Microelectronic and Infineon.
Two Very Different Business Models
That split matters for device makers. A low-cost BLE chip can be enough for a simple beacon or a basic remote control. Products that need long battery life, strong security, regulatory certification for medical use, or on-chip processing tend to pay more for the silicon, and that is the segment where differentiation is growing fastest.
As TSR frames it, the contest in IoT Bluetooth is less about who can ship the cheapest radio and more about what else the chip can do.
Positioning: Channel Sounding Moves Into Products
The most important new capability is precise ranging. Bluetooth Core 6.0, published in 2024, introduced Bluetooth Channel Sounding, a ranging feature that lets two devices measure the distance between them far more accurately than older signal-strength methods.
Older Bluetooth location relied mainly on received signal strength, which is easily thrown off by walls, bodies and interference. Channel Sounding combines phase-based ranging with round-trip timing to estimate distance, and adds protections designed to make it harder to fake proximity.
Where Ranging Matters
TSR expects commercial deployment of Channel Sounding to begin in 2026. The first use cases are ones where knowing how close something is really matters:
· Digital keys for cars and buildings, where a door should unlock only when the authorised phone or fob is genuinely nearby.
· Item finding, where a tag can guide a user to a lost object with greater accuracy.
· Access control in offices, hotels and homes.
· Asset tracking indoors, where location needs to be finer than a whole room or zone.
For the car market in particular, this puts Bluetooth into more direct competition with ultra-wideband (UWB), which has so far been the preferred technology for high-accuracy digital keys. ABI Research expects UWB device shipments to grow from 597 million in 2026 to 1.18 billion by 2030, so the two technologies are likely to coexist, with Channel Sounding offering a lower-cost option where a separate UWB radio is hard to justify.
Edge AI Moves Onto The Radio Chip
The second big shift is intelligence. Wireless SoC vendors are adding machine learning capability directly to their chips, ranging from tiny models that run on the main processor to dedicated neural processing units (NPUs). The aim is to let a sensor or wearable interpret data locally, and transmit only what matters.
"Wireless connectivity is no longer just about linking devices, it is increasingly about enabling scalable edge AI," Andrew Zignani, Senior Research Director at ABI Research, said in the firm's July forecast.
What That Looks Like In Silicon
Nordic offers a clear example. At CES 2026, the company introduced the nRF54LM20B SoC, which adds its Axon NPU to a Bluetooth LE chip, alongside a set of very small Neuton machine learning models, typically under 5 KB, and a development tool it calls the Nordic Edge AI Lab. Nordic targets tasks such as sound classification, keyword spotting, gesture recognition and anomaly detection.
"Edge AI is no longer optional – it's the only way to deliver safety, privacy, and sustainability at scale," Nordic chief executive Vegard Wollan said at launch. Nordic's performance and efficiency claims against rivals are its own, and real-world results will depend on the model and workload.
The broader trend is the same one we have seen in larger devices. Our piece on the cellular module becoming a computer traced how connectivity hardware is taking on processing work, and our look at how physical AI turns sensing into action explains why local decisions matter for IoT systems that need to respond quickly.
Why Local Processing Helps Battery Devices
For small battery-powered products, the radio is often the biggest energy drain. Sending raw sensor data to a phone or gateway all day consumes power. A chip that can recognise a pattern locally, such as an abnormal vibration or a specific spoken word, only needs to wake the radio when something relevant happens. That can extend battery life and reduce the amount of personal data leaving the device.
Ambient IoT: The Next Frontier, Still Early
The third trend is further out. Ambient IoT refers to devices that run on harvested energy, from light, radio waves, motion or heat, instead of batteries. The goal is tags and sensors cheap and small enough to attach to packages, pallets and products by the billion.
TSR sees Bluetooth-based ambient IoT still in the standardisation stage, with broader commercial use expected around 2028. The Bluetooth SIG has been working on how Bluetooth fits ambient IoT, while other industry groups and standards bodies, including 3GPP for cellular, are developing competing and complementary approaches.
We looked at the practical side of this in our profile of Sensolus tracking assets without power. For Bluetooth chip vendors, the challenge is to push power consumption low enough that energy harvesting can reliably run the radio, without making the tags too costly.
What This Means For Device Makers
For product teams choosing Bluetooth silicon over the next few years, the shift in the market brings a few practical questions:
· Does the product need ranging, and is Channel Sounding enough, or does it still need UWB?
· Would on-device machine learning cut radio use, battery size or cloud costs enough to justify a more capable chip?
· How long does the product need to stay in service, and does the chip vendor provide long-term software and security support?
· Is the lowest unit price the right trade-off if the product needs medical, automotive or security certification?
The answers vary widely by product, which is one reason the market is splitting between low-cost volume suppliers and vendors selling more capable, higher-priced SoCs.
The Bottom Line
Bluetooth's growth story is no longer only about earbuds. TSR's data points to a market where audio matures, and IoT uses such as health monitoring, tracking, smart home and digital keys provide most of the new demand.
The chips themselves are changing with that shift. Precise ranging through Channel Sounding and built-in machine learning are turning Bluetooth from a simple cable replacement into a sensing and decision-making platform. Energy-harvesting ambient IoT could extend that further, although its commercial impact is still a few years away.
