Innatera and the EIC Accelerator: Brain-Inspired Chips for Sensors
Innatera, a 2018 spin-out of Delft University of Technology registered in Rijswijk, designs "neuromorphic" microcontrollers that run spiking neural networks, which process sensor data in bursts like brain cells, so battery-powered devices can recognise sounds, gestures or presence locally at very low power. It was selected in the EIC Accelerator's January 2023 cut-off for blended finance and received an EU grant of €2.49 million for "Always-On Intelligent Sensing with the Spiking Neural Processor", a €3.56 million project that ran from May 2023 to April 2025, according to CORDIS. The goal was a silicon prototype of a new analogue-mixed-signal computing fabric.
The project is closed. Its final report states that Innatera gathered requirements with early customers, designed new computing and sensor-interface blocks, built and characterised a silicon prototype, and ran a pilot application on it, ending with "a commercially exploitable architecture", according to the CORDIS reporting page. On 21 May 2025, a month after the project ended, Innatera launched Pulsar, "its first commercially available microcontroller", which it says delivers "up to 100X lower latency and 500X lower energy consumption than conventional AI processors", according to its launch announcement.
Company file
- Legal entity
- Innatera Nanosystems BV, Rijswijk
- Co-founder and CEO
- Sumeet Kumar
- Origin
- Delft University of Technology spin-out, 2018
- Product
- Pulsar neuromorphic microcontroller; Talamo SDK
- Chip size
- 2.8 × 2.6 mm package
- Customers named
- Joya (ODM); VLSI Expert (training)
- EIC funding
- Blended finance; €2.49M grant
Where Innatera Stands Now
At CES in January 2026 Innatera showed customer devices using Pulsar and said it had begun work with Joya, a global original design manufacturer, to put Pulsar into lifestyle, IoT and smart-home products, according to PR Newswire. Joya's founder said the collaboration "is just beginning". Pulsar combines analogue and digital spiking-network engines, a RISC-V processor and accelerators for conventional neural networks, typically running below 1 mW, according to XPU.pub.
The record
From university spin-out to a chip on sale
Innatera built a silicon prototype during its EIC project and launched Pulsar, its first commercially available neuromorphic microcontroller, a month after the project ended.
- 2018Innatera foundedSpin-out of Delft University of Technology
- Jan 2023EIC Accelerator cut-offSpiking Neural Processorblended finance
- May 2023Project starts€2.49M EU grant
- Jun 2024Series A completed$21M
- Apr 2025Project endsSilicon prototype characterised
- May 2025Pulsar launchedComputex 2025
- Jan 2026CES 2026 deploymentsWork begun with ODM Joya
CORDIS, Innatera, XPU.pub and PR Newswire.
"It marks the moment that our brain-inspired technology becomes ready for mass-market deployment."
Sumeet Kumar, co-founder and CEO, Innatera, May 2025The Technology: Chips That Spike
Most sensor data carries nothing of interest; the final report puts the share at over 95% in common uses. Sending it all to a processor or the cloud wastes energy. Spiking neural networks only compute when an input changes, which suits always-on sensors such as radar, microphones and infrared. Innatera's chip handles these workloads next to the sensor, so devices running on small batteries can make decisions locally.
The claims
Pulsar against conventional AI processors, as claimed
Innatera says Pulsar delivers up to 100 times lower latency and 500 times lower energy use than conventional AI processors. These are the company's own figures.
Innatera, Pulsar launch, May 2025. Multiple of improvement claimed.
What the EIC Award Promised, and What the Record Shows
The award promised a silicon prototype of a new near-sensor computing fabric. The final report documents that prototype, and the company followed it with a commercial product. The link between the two is not spelled out publicly: Pulsar's launch announcement does not state which of its blocks came from the EIC project. The performance claims are the company's own.
- Market requirementsDeliveredWith early customers.
- Silicon prototypeDeliveredCharacterised in year two.
- Commercial productDeliveredPulsar, May 2025.
- Volume salesUnverifiedNo figures published.
The Money
Innatera completed a $21 million Series A in June 2024 on top of $6 million of seed funding, XPU.pub reports. Whether the EIC Fund invested under the blended finance award, and on what terms, has not been disclosed by the company, and its revenue and losses could not be checked from free sources.
The funding
Disclosed funding
The Series A was more than eight times the EIC grant. Dollar rounds are shown in US dollars.
XPU.pub and CORDIS. Values in millions, in the currency stated.
The project budget
Who paid for the €3.56M project
The EU covered 70% of the project's cost.
CORDIS, project 190123060. Values in € millions.
What Cannot Be Checked
Questions for the EIC
Whether the EIC Fund invested, how many Pulsar chips have been sold, and which EIC-funded blocks are in the commercial product are not public.
Why Europe Should Care
The final report argues that processing data next to billions of sensors, instead of in the cloud, could save millions of tonnes of CO2 a year. The EIC funded the prototype of a Dutch spin-out's chip, and the company has since put a commercial chip on sale. Whether it becomes a volume product depends on device makers such as Joya, whose collaboration its founder called "just beginning".
What the award promised
A silicon prototype of a new always-on, near-sensor computing architecture.
What the record shows
A characterised prototype, a commercial chip launched weeks later and first device makers adopting it.
The Verdict
From prototype to product
Innatera delivered its EIC project and brought a chip to market. How well it sells will decide whether the bet pays off.
Innatera and the EIC have not been asked for comment for this article. It is based entirely on public records and published statements, linked throughout.
Industry
Electronics and semiconductorsSemiconductors
- Focus
- Processors
- Technologies
- Hardware acceleration
- Low-power electronics
- Keywords
- Spiking neural networks
- Edge AI
- Smart home devices
- Always-on sensing
- Neuromorphic computing