Gani
Jusuf

Picture the inside of a vehicle approaching full autonomy. Not the consumer pitch - the engineering reality. Cameras pointing in six directions. Radar sweeping 360 degrees. LiDAR pulsing millions of points per second. All of that sensor data needs to travel from the roof and bumpers to a central compute unit, fast enough, reliably enough, and securely enough that the car doesn't kill anyone. The wiring harness alone in a modern premium vehicle weighs over 50 kilograms. The data pipes are the bottleneck no one in the press talks about.

Gani Jusuf has been thinking about that bottleneck since before most people knew it existed. In February 2018, after nearly 15 years as Vice President of Engineering at Marvell Technology - one of the world's largest semiconductor companies - he walked away from a large-cap VP chair and co-founded AXONNE. The company's mission: build the chips that move sensor data across a vehicle at multi-gigabit speeds, over the cheapest, lightest, most available medium possible. A single twisted-pair cable.

This is not a software story. It is a physics and silicon story. The kind that takes decades of engineering intuition to attempt, and where the stakes are literal road safety.

Jusuf did not dabble at UC Berkeley. He went back three times - earning a B.S., then an M.S., then a Ph.D., all in Electrical Engineering and Computer Science. His doctoral research touched analog IC design, including work on automatic synthesis of CMOS analog-to-digital converters - the kind of paper that ends up cited 254 times by other engineers who later build on the same ideas.

From Berkeley he went to Hewlett-Packard, then to Marvell Semiconductor as Director of Engineering, and eventually spent 14 years as VP of Engineering for the Communications and Consumer Business Group at Marvell Technology Group. Marvell is not a household name, but its chips are in most of the routers, storage drives, and network switches you've touched in the past decade. Running engineering for that division means you have solved hard-bandwidth problems at industrial scale, with real customers measuring you in sub-nanosecond increments.

Along the way, in 2014, Jusuf joined the board of PowerbyProxi, a New Zealand startup working on wireless power technology. He called it "still developing" with "enormous potential." Apple acquired PowerbyProxi in 2017. Jusuf has a record of early conviction on emerging connectivity technology that later becomes obvious.

AXONNE's founding premise is straightforward once you understand the problem: the automotive industry is transitioning from dedicated, point-to-point wired connections to Ethernet-based networking. That transition has to happen at multi-gigabit speeds, over the same twisted-pair copper wires already in vehicles (not fiber, not coax, not new exotic mediums), with automotive-grade reliability ratings (AEC-Q100 Grade 2), and with security encryption baked in at the hardware level - MACsec, to prevent the kind of cyberattacks that could theoretically take control of a vehicle.

The flagship product is the Cyton PHY - a transceiver chip implementing IEEE 802.3ch Multi-Gigabit Automotive Ethernet. The Cyton handles 1Gbps to 10Gbps over a single unshielded twisted pair or coax cable. In January 2023, AXONNE and Leopard Imaging demonstrated the world's first multi-gigabit automotive Ethernet camera development module at DesignCon. That demo matters because cameras are the dominant sensor type in ADAS systems - and their resolution demands keep climbing.

The AXONNE team is small - roughly 27 people - spread across San Jose and Sunnyvale in California, plus offices in Karlsruhe, Hamburg, and Bucharest. The European footprint is not accidental. BMW, Mercedes-Benz, and the rest of the German OEM ecosystem are among the world's most aggressive adopters of next-generation vehicle architectures. You don't open offices in Karlsruhe unless you're having serious conversations with Stuttgart and Munich.

In May 2020, Intel Capital deployed part of a $132 million investment across 11 disruptive technology startups. AXONNE was on that list. Intel's automotive ambitions - via Mobileye and its own automotive compute division - made the thesis obvious: you need the chips that move the data before you can process it. The Cyton PHY is that foundation layer.

By early 2026, AXONNE was appearing at AEC2026 in Munich, demonstrating advanced camera module technology to European automotive OEMs. The industry calendar placement signals a company that has moved past proof-of-concept and is having procurement conversations.

SerDes - serializer/deserializer - is the fundamental technology behind AXONNE's products. The idea: take parallel data, serialize it into a high-speed serial stream, transmit it over a single wire pair, then deserialize it on the other end. In theory, simple. In practice, getting 10 gigabits to travel reliably across a vehicle wiring harness - which is a hostile electromagnetic environment full of noise from motors, inverters, and power electronics - requires a level of analog RF and mixed-signal design expertise that takes decades to develop.

Jusuf and his co-founder William Lo came from Marvell, a company that built exactly this expertise for enterprise networking (the technology that made gigabit Ethernet ubiquitous in data centers and offices). The automotive translation is not trivial: the temperature ranges are wider, the vibration tolerances are tighter, the certification requirements (AEC-Q100, ISO 26262 functional safety) are demanding, and the OEM qualification cycles run years, not months.

AXONNE's founding team claims to have "first brought the realization of Automotive 1000BASE-T1 Ethernet into the market" - meaning they built this product category before starting AXONNE. That's a significant credential. It means the Cyton PHY is not a first attempt. It is a second-generation product built by people who already shipped the first generation somewhere else.