Seventy-eight million batteries. Every single day.
That's the number of IoT power cells hitting landfills and incinerators in 2025, per IEEE Transmitter—a torrent of lithium and heavy metals leaching into soil and groundwater at a scale that regulators are finally taking seriously. Europe's new battery mandates kick in through 2027. California just slapped a retail surcharge on devices with embedded batteries. The compliance costs are piling up.
But here's the thing: the sheer volume of waste is also creating a market opening that a handful of chipmakers think they can exploit. What if millions of those connected sensors, tags, and monitors didn't need batteries at all?
It's a radical proposition that's been stuck in university labs for the better part of two decades. Now it's shipping silicon. And some very serious money—venture capital, corporate R&D, and European industrial policy funds—is betting it's about to hit commercial scale.
The Chip That Runs on Thin Air
At the forefront sits ONiO.zero, a microcontroller from a Norwegian startup staffed by veterans of Atmel, Nordic Semiconductor, and Novelda. The pitch is deceptively simple: a chip that consumes just 22 microwatts per megahertz—roughly five times more efficient than leading ARM-based alternatives—and can cold-start on less than one microwatt of harvested ambient energy.
What does that actually mean? The device scavenges power from whatever's lying around: RF signals bouncing through a warehouse, indoor lighting, temperature gradients across a pipe, even vibration. It's not vaporware. ONiO raised €5 million last June from node.vc and Maki.vc, with backing from the European Innovation Council and a Danish pension fund. They're ramping production now, targeting a market that multiple analyst houses peg between $12 billion and $15 billion by 2030.
The founders—Kjetil Meisal, Runar Finanger, and Vemund Kval Bakken—argue that the ARM ecosystem has hit diminishing returns on power efficiency. Their solution integrates what used to require three or four separate chips: a RISC-V microcontroller, Bluetooth 5.4 and IEEE 802.15.4 radios, 64KB of ultra-low-power flash, and a multi-input energy-harvesting power management IC. All on one die.
That integration, ONiO claims, shaves 60% to 80% off bill-of-materials costs versus pairing a discrete ultra-low-power MCU with an off-the-shelf PMIC. Bold numbers. Whether they hold up at volume remains to be seen.
A Market That's No Longer Hypothetical
The category analysts call "ambient IoT"—devices powered entirely or primarily by harvested energy rather than primary batteries—is moving fast. ABI Research forecasts 1.1 billion ambient IoT device shipments by 2030, with photovoltaic harvesting accounting for 57% of deployments, RF energy for 36%, and piezoelectric and thermoelectric sources picking up the rest.
Multiple trackers converge on similar trajectories. Mordor Intelligence pegs the ultra-low-power microcontroller market at $8.22 billion this year, growing to $12.69 billion by 2030—a 9.07% compound annual growth rate. MarketsandMarkets goes slightly higher: $9.78 billion in 2025, climbing to $15.27 billion by decade's end. The energy-harvesting systems market itself—PMICs, supercapacitors, rectifier circuits—is projected to nearly triple from around $850 million to $2.2 billion by 2034.
The technology has escaped the lab. Wiliot's battery-free Bluetooth tags are being manufactured at scale by Avery Dennison for supply-chain tracking. Samsung ships millions of TV remotes annually that harvest energy from indoor light and ambient Wi-Fi signals, storing it in supercapacitors. EnOcean's battery-free building sensors have been retrofitted into thousands of European properties, including Dresden's historic Bellevue Hotel, where wireless switches eliminated the need to tear through century-old walls to run new wiring.
Perhaps most telling: Everactive's batteryless steam-trap monitors, deployed in a pilot at the University of Virginia, documented a 90% reduction in steam energy loss with quantified CO₂ and cost savings. That's the sort of real-world proof point that makes procurement officers pay attention.
Why Now? Three Converging Forces
Regulation, economics, and silicon innovation.
Start with regulation. Europe's Battery Regulation—rolling out in phases through 2027—mandates that portable batteries in consumer devices must be user-replaceable. Collection targets: 63% by 2027, 73% by 2030. Digital battery passports become mandatory for certain categories starting February 18, 2027. California's new Covered Battery-Embedded Waste Recycling Fee took effect January 1, imposing a 1.5% retail surcharge (capped at $15) on devices with embedded batteries. Retailers are now responsible for collection logistics.
For product managers designing the next generation of IoT endpoints, these aren't abstract compliance exercises. They're line items that make battery-free designs suddenly competitive on total cost of ownership.
The economics extend beyond regulatory penalties. Consider: a battery-powered sensor node might cost $2 in components but require field visits every 18 to 36 months for replacement. Multiply that across thousands or tens of thousands of deployed devices, and the labor cost obliterates the bill of materials. Battery-free designs eliminate that maintenance cycle entirely—assuming the energy budget works.
Which brings us to the silicon breakthroughs.
ONiO.zero's integrated architecture handles what used to demand separate components. The chip's internal RF rectifier operates across ISM and GSM bands (800/900/1800/1900/2400 MHz). The photovoltaic interface can cold-start from 250 millivolts. Thermal and piezoelectric inputs accept 0.25 to 1.0 volts. It couples directly to capacitors, supercapacitors, or small secondary cells for energy buffering.
That cold-start spec—under one microwatt—undercuts most external PMICs, which typically need three to thirty microwatts to bootstrap. It's a meaningful edge, though competitors aren't standing still.
The Competitive Scrum
The landscape is crowded and fragmented.
Atmosic's ATM33e and ATM34e SoCs offer on-chip RF energy harvesting alongside Bluetooth 5.4 and 802.15.4 radios, though they stick with ARM Cortex-M33F cores rather than RISC-V. Renesas' RE family uses a proprietary silicon-on-thin-buried-oxide process to hit 25µA/MHz in active mode (12µA/MHz with an external DC-DC converter) and includes energy-harvesting control—but lacks integrated radios. Ambiq's Apollo4, often cited for efficiency at 5µA/MHz, doesn't integrate harvesting at all. Designers pair it with discrete PMICs.
Nordic's nRF54L series and STMicroelectronics' STM32U5 family represent the incumbent ARM approach: excellent power efficiency (Nordic claims around 20µA/MHz processing efficiency), mature toolchains, and broad ecosystem support. No native energy harvesting, but they're battle-tested in hundreds of commercial products.
ONiO, by contrast, is a startup founded in 2018. Their bet: the ARM ecosystem has hit a power efficiency ceiling, and the industry needs a fundamentally lower baseline to make battery-free operation viable beyond niche use cases.
Real-World Deployment: Where the Rubber Meets the Road
Theory is one thing. Shipping products, another entirely.
Wiliot's battery-free Bluetooth pixels—now in production with Avery Dennison and embedded in Tageos' EOS-654 BLE G3 inlays—are being piloted by retailers including Walmart for pallet and case-level tracking. The pixels harvest RF energy from Wi-Fi access points and communicate via Bluetooth backscatter. The economics work because battery replacement across tens of thousands of tags simply isn't scalable.
In building automation, EnOcean's switches and sensors have carved out a foothold in heritage retrofits where running new wiring is prohibitively expensive or outright impossible. Dresden's Bellevue Hotel installation eliminated miles of potential conduit runs through protected historic structures. The switches harvest energy from the mechanical press of a finger—piezoelectric transduction—and transmit wirelessly.
Everactive's steam-trap monitors at UVA run entirely on harvested thermal gradients from the very pipes they're monitoring. It's an elegant match of energy source to application: the thing you're measuring provides the power to measure it. The university documented savings that penciled out to a sub-12-month payback on sensor deployment costs.
ONiO has partnered with Vinatech to integrate supercapacitors into reference designs for buffering intermittent harvested energy. They're working with BeFC to co-develop paper-based biofuel cell solutions for disposable medical sensors—glucose monitors, temperature patches. Early customer pilots include batteryless electronic shelf labels, soil moisture sensors, and air-quality monitors.
Whether any of this scales beyond pilots is the open question.
The Path from Niche to Mainstream
Three variables will determine whether ambient IoT becomes ubiquitous or remains confined to edge cases: standards, software tooling, and cost curves.
Standards momentum is building. 3GPP Release 19, expected to stabilize next year, includes an Ambient IoT study item targeting "Device 1" class endpoints that consume around one microwatt peak power and communicate via backscatter on licensed cellular bands. Bluetooth 5.4 now supports coded PHYs and direction finding suitable for low-power asset tags. The Ambient IoT Alliance—founded by Wiliot, Qualcomm, Intel, Infineon, and PepsiCo—is pushing cross-vendor interoperability.
Matter and Thread, both built on 802.15.4, are gaining traction for smart-home devices that increasingly incorporate energy-harvesting designs. The ecosystem is coalescing, if unevenly.
Software tooling? That's where things get messy.
Ambient-powered devices experience intermittent operation by definition. They wake up, sense, compute, transmit, then go dormant until enough energy accumulates for the next cycle. This breaks assumptions baked into most embedded operating systems. Academic research on checkpointing, non-volatile processor state, and approximate computing frameworks—projects like Approxify and PEARL—is maturing. But commercial RTOS support lags badly.
ONiO runs bare-metal without an OS, which simplifies the power budget but limits application complexity. As the category scales, expect more robust middleware and compiler toolchains that make intermittent computing less exotic. It's coming, but it's not here yet.
Cost curves will ultimately determine adoption velocity. Wiliot's partnership with Avery Dennison aims to drive per-unit costs of battery-free BLE labels below those of active BLE beacons (typically $5 to $15 each) and eventually approach passive UHF RFID price points—under ten cents at high volume.
Whether that's achievable depends on yield, packaging innovation, and antenna integration. For microcontrollers like ONiO.zero, the question is less about absolute silicon price than total system cost. If integrating the PMIC on-die saves two external components and simplifies board layout, the value proposition holds even at price parity with discrete solutions.
Where It Works (And Where It Doesn't)
The opportunities are clearest in applications where battery replacement is impractical or prohibitively expensive: sensors embedded in concrete infrastructure, tags on perishable goods moving through cold chains, wearables in medical settings, industrial monitors in hazardous or hard-to-reach environments.
The challenges? Equally clear.
Energy harvesting is inherently unpredictable. Indoor lighting varies. RF environments shift. Metal enclosures and wet conditions starve harvesters. Designs must gracefully handle power failures, which means more sophisticated state management and sometimes hybrid architectures that pair harvesting with small rechargeable buffers as backstops.
There are also application mismatches. High-duty-cycle devices—anything transmitting continuously or running computationally intensive workloads—won't run on harvested energy alone. Physics doesn't care about your product roadmap.
But for a growing slice of the IoT universe—low-duty-cycle sensing, periodic check-ins, event-triggered alerts—the math is starting to work.
The 78-Million-Battery Question
For hardware engineers and product managers in the IoT space, the calculus is shifting. Battery-free isn't a panacea. It won't replace lithium-ion in your smartphone or electric vehicle. But it's no longer a curiosity, either. It's a design option backed by shipping silicon, tightening regulations, and mounting evidence that maintenance-free autonomy can pencil out.
The seventy-eight million batteries discarded daily won't disappear overnight. That would require a transformation of consumer electronics far beyond IoT endpoints. But the trajectory is bending, perhaps more than the industry expected even three years ago.
And that's the real story—not the elimination of batteries, but the emergence of a credible alternative that's moving from lab demos to commercial deployments at a velocity that's catching incumbents off guard. Whether ONiO and its cohort of energy-harvesting chipmakers capture significant market share or get steamrolled by ARM licensees adding similar features remains to be seen.
Either way, the 78-million-battery problem just got its first serious answer. How the market responds will determine whether ambient IoT becomes the next platform shift—or another overhyped technology that couldn't quite escape its niche.
