A chip design team out of Imperial College London has raised $5M to sell its chips to the companies building brain implants, rather than build an implant of its own. MintNeuro, the trading name of Mint Neurotechnologies, announced the round on 7 October, led by the London fund Odyssey Ventures with 33East of Cyprus, the Czech fund JSK Investments and Alumni Ventures joining, and Empirical Ventures and Jumpspace Ventures returning.
Neither the company nor its backers put a label on the round, so there is no seed or Series A stage attached to it, and no valuation was disclosed. MintNeuro was founded in 2022 out of the Next Generation Neural Interfaces lab in Imperial's electrical engineering department, and its chips are designed to be bought in, not built in house by its customers.
Dorian Haci, a former postdoctoral researcher at Imperial, is chief executive and co-founded the company with Timothy Constandinou, the professor of bioelectronics who runs that lab and serves as chief technology officer. Tim Denison of the University of Oxford chairs the board, and Andrew Jackson of Newcastle University is chief scientific officer.
What Has To Fit Inside The Skull
A company building an implanted neural device today usually assembles it from off-the-shelf parts, and the extra circuitry needed to make those parts talk to each other adds bulk and drains power. Safety rules add more components still, among them parts that survive physical impact and parts that cap stimulation current at levels that cannot injure tissue. Ingenia, the Royal Academy of Engineering's magazine, set out that problem in May 2025.
Custom silicon removes the glue logic, and MintNeuro's answer is a modular family of blocks covering signal recording, electrical stimulation with safety limits built in, on-chip data reduction, power management and wireless transfer. Each chip measures roughly 2mm by 2mm and comes off lines at TSMC, the Taiwanese contract manufacturer, built at 65 nanometres and above. Those are mature manufacturing processes rather than the cutting edge.
Stimulating nerve tissue needs voltages and currents the newest manufacturing processes cannot supply, which is the reason Haci gives for staying on older ones. An implant is constrained by heat and power rather than by raw speed. Industry guidance holds an implant to a temperature rise of one degree Celsius in surrounding tissue, which puts a hard ceiling on how much electronics can be packed in.
A paper in the journal npj Biomedical Innovations put numbers on the bottleneck in August 2026, describing a system that reads 5,376 electrode channels at once while drawing about 42 microwatts on each, or millionths of a watt. Shipping that data out of the body accounted for most of the power the system used, which is the part MintNeuro's on-chip data reduction is aimed at. The researchers also noted that their own chip sits outside the body, where air carries the heat away.
One Named Customer So Far
Motif Neurotech, a Houston company developing a stimulation device the size of a blueberry, signed a multi-year supply agreement with MintNeuro in April 2026 covering Motif's animal and early human studies and the large final trial that precedes a regulatory submission. A MintNeuro sensing chip is already inside those devices in pre-clinical work, and the two had worked together for more than a year before saying so publicly. Motif won approval in April 2026 for a US clinical trial in treatment-resistant depression.
Amber Therapeutics, a UK company developing an implant for mixed urinary incontinence, is the only other customer MintNeuro names, and it says the pair have reached integration milestones. They hold a joint grant from Innovate UK, the government's innovation funding agency, to develop a nerve stimulation chipset, and no commercial supply agreement between them has been announced.
Beyond those, the announcement says that several neural device companies are evaluating or integrating the technology, and that the money will expand delivery of pre-release silicon to partners. None of those companies is named, and the phrase pre-release silicon marks the product as not yet generally available. Haci framed the step as moving "from technology validation to commercial scale-up".
Others Already Sell Neuro Chips
Cactus Semiconductor, part of Cirtec Medical and based in Chandler, Arizona, has been designing low-power chips for implanted devices for years, and its catalogue lists integrated circuits for implanted drug delivery and for implanted nerve stimulators. Sigenics, based in Chicago, Illinois, designs custom chips that handle analogue signals and digital ones together, and cites neural stimulation and neural signal recording among its specialisms.
Intan Technologies sells neural amplifier chips off the shelf to laboratories in more than 50 countries, with one part combining stimulation and amplification. Intan makes no claim that its chips are meant for human implantation, which is the distinction MintNeuro is working with.
The companies with the deepest pockets design their own silicon in house. Neuralink, Blackrock Neurotech and Paradromics all build their chips themselves. They are the customers MintNeuro would most like to win and the reason it may not need to be won over, since each has already paid for the capability in house. MintNeuro's case is that a reusable, pre-validated platform built to implant standards beats both bespoke design work and research-grade parts, and that case rests on positioning rather than on any published measurement.
$5M Against $250M Rounds
Precision Neuroscience raised $250M in September 2026, Neuralink took $650M in June 2025 at a reported valuation near $9B, and Synchron closed $200M in November 2025. Those companies build the implants; MintNeuro sells a component into them, which is a different layer of the same industry and a different scale of capital requirement.
Grant money has done more of the work than equity so far. MintNeuro raised £1M in November 2024, led by Empirical Ventures with Parkwalk Advisors, Imperial College London, Plug and Play and Excellis Holding alongside. It also sits inside three consortium projects together worth £17M, funded under the UK Advanced Research and Invention Agency's £69M neurotechnology programme, and its own share of that money has never been published.
MIT Technology Review reported in April 2025 that fewer than 100 people worldwide have ever lived with an implanted brain-computer interface, meaning a device that reads brain activity and turns it into commands. A researcher quoted in that report counted 71 patients who had controlled a computer with their neurons across 26 years, and around a quarter of implant recipients appear nowhere in the medical literature at all.
Precision Neuroscience holds a US clearance for its electrode array limited to 30 days of use, which permits sale for that narrow purpose rather than general approval. Neuralink, Synchron and Onward Medical hold breakthrough designations, which speed up a regulator's review without granting anything, and Paradromics implanted its first long-term patient in June 2026. No implanted brain-computer interface has full market approval anywhere in the world.
What The Chips Cannot Show Yet
MintNeuro publishes no channel count, no power figure per channel and no noise or delay numbers, and a journalist who interviewed Haci in April 2026 recorded them as undisclosed. The company says its approach allows implants up to 1,000 times smaller than current devices, a figure nobody outside it has measured. That leaves the Motif agreement as the firmest public evidence that the chips work.
Specialised chips displacing bolted-together parts is a pattern that has already played out elsewhere in electronics, where Bluetooth parts quietly absorbed the jobs once done by separate components. Whether it plays out the same way inside the skull depends on device makers choosing to buy rather than design, and none of them beyond Motif has said publicly that it will.
