A quiet operator in a noisy moment for AI.
The chip is the size of a fingernail. It listens to a sentence, transcribes it, and uses about as much electricity as a wristwatch. That is the demo Applied Brain Research has been running in its Waterloo lab in 2026. The CEO who is preparing to ship it has spent his entire working life inside the kind of silicon that, until recently, nobody wrote stories about.
Kevin Conley runs Applied Brain Research from a corner of two geographies. His company files from 200 University Avenue West in Waterloo, Ontario. He files his tax returns from Los Gatos, California. The arrangement is fitting for someone whose career has been a bridge between the things engineers build and the markets that buy them.
The long apprenticeship
Conley spent more than two decades at SanDisk, an unusual tenure in an industry where the average executive plays musical chairs every three years. He arrived as a systems engineer, climbed through a series of leadership roles, ran Client Storage Solutions as Senior Vice President and General Manager, and left as Senior Vice President and Chief Technology Officer in June 2016. Somewhere along the way he was credited as an inventor or co-inventor on more than ninety patents covering non-volatile memory architecture and management. That is a number larger than the entire current headcount of his new employer.
Between SanDisk chapters there was a year at Corsair as Vice President of Engineering, expanding the company's PC components business into high-performance SSDs. Then in 2017 he took a Board seat at Everspin Technologies, the Arizona company commercializing MRAM - magnetoresistive random-access memory. Six months later he was its President and CEO. Boards do not usually do this. The signal was: we need an operator who already understands the product.
He ran Everspin until early 2021, navigating the specialty memory market through the kind of transition that public-company CEOs rarely talk about in the abstract: pruning, refocusing, surviving long enough for the technology to find its market. The Wall Street Transcript called him a semiconductor storage veteran. The phrase undersells him. He is also an inventor and an operator. The combination is rare.
The pivot
Applied Brain Research was founded in 2012 by alumni of the Computational Neuroscience Research Group at the University of Waterloo. Its co-founders are largely Ph.D. researchers - Chris Eliasmith, Travis DeWolf, Daniel Rasmusen, Trevor Bekolay, Xuan Choo - and for a decade the company was best known in academic and Nengo simulator circles, not press releases. The work was brain-inspired computing: how do you build silicon that thinks the way neurons do, in spikes and time, rather than the way GPUs do, in matrix multiplications and watts?
In late 2023 the company answered a different question: who runs the business that turns this research into a product? They picked Conley.
"I am thrilled to join ABR as CEO and embark on this exciting journey. In an era of rapid technological evolution, ABR stands at the cutting edge of AI solution innovation, and I am committed to bringing its game changing technology to the Artificial Intelligence of Things market."- Kevin Conley
Co-founder and now CTO Chris Eliasmith framed the appointment in plain terms: "ABR's commitment to delivering breakthrough solutions aligns seamlessly with Kevin's proven ability to drive product development and market expansion." Translation: the founders had built the science. They needed a closer.
The wager
The thesis at ABR is small, specific, and increasingly contrarian. The dominant story of AI in 2026 is bigger - bigger models, bigger clusters, bigger power bills. Conley's company is selling the opposite. A time-series AI processor designed for sensor data. Spiking neural networks. State-space neural networks. Legendre memory units. On-chip non-volatile memory. The numbers ABR publicizes describe what happens when you stop trying to be a GPU and start trying to be a brain: 100x lower power than other high-functionality edge AI hardware, speech recognition running at 35 milliwatts with 120 milliseconds of latency.
The A0 silicon is up and running. Full production is targeted for the second quarter of 2026. If you have read enough press releases to be skeptical of those words, you will recognize the difference between a research lab and a man who has shipped products in volume before. ABR has the lab. Conley brings the second part.
The contrast
It is worth pausing on the strangeness of the resume. The man who helped scale NAND flash into a generation of laptops, phones, and SSDs is now backing a company whose pitch is that those very devices waste energy when they listen to you. The CEO who spent years making memory denser and cheaper is now selling memory that is, by design, on the chip and in the model itself.
This is not a reversal. It is a continuation. Each chapter of Conley's career has been about a substrate that nobody outside the industry thinks about - and a stubborn belief that the substrate is where the next decade is won. NAND was a substrate. MRAM was a substrate. Spiking, brain-inspired silicon is a substrate. He has spent thirty years optimizing the parts of computing that the rest of us take for granted.
Off-page
The biographical material is sparse in the way it tends to be for operators rather than founders. He earned a Bachelor of Science in Electrical Engineering and a Master of Science in Computer Engineering, both from Santa Clara University. He is an alumnus of the Stanford Executive Program. He keeps a low-key LinkedIn presence under the handle kconley. The Twitter handle on file for the company is @abr_inc, not his own. He does not appear to spend his evenings on the conference panel circuit. He is, in short, the kind of CEO who shows up in the press release announcing the silicon, and then again in the press release announcing the design wins, and not much in between.
What it adds up to
If the next phase of AI belongs to the devices that listen, watch, and respond at the edge - earbuds, wearables, industrial sensors, medical monitors, in-cabin voice in cars - somebody has to build the chips that make that arithmetic affordable. The companies that win that race will not be the ones with the most flops. They will be the ones with the lowest milliwatts.
Kevin Conley is betting his fourth act on that sentence. The interesting thing about him is not that he is making the bet. It is that he is the kind of person who, having spent ninety patents and three decades inside the box, is now arguing for the cleverness of stepping outside it.