Ethernovia

We'll be in Gothenburg for the IEEE Ethernet & IP @ Automotive Technology Day, 14–15 October. We've booked a private meeting room at Gothia Towers right next to Svenska Mässan. It's away from the conference floor so there's time to properly dig into your architecture, requirements and roadmap. Our team will be showcasing: • Our latest automotive Ethernet PHY silicon, from 1G up to 10G • ForeSight Diagnostics™ and how it solves key issues in today's networks • A range of demo's covering our product portfolio to enable the nervous system for intelligent machines • Practical ways to take cost and complexity out of the network Meetings are by appointment, please reach out to Nicolae Gaidibadi to schedule yours now. See you in Gothenburg! #AutomotiveEthernet #IEEE #Gothenburg #SoftwareDefinedVehicle

Last week Arm brought together more than 80 companies spanning silicon, sensors, safety software and cloud infrastructure into a new physical AI ecosystem called Arm Total Design. NXP Semiconductors posted about sensor fusion and coordinated understanding the same week. Everyone is converging on the same idea: physical AI systems need to sense, decide, communicate and act as one coordinated system rather than a pile of separate components. The compute and safety frameworks for that vision are getting built fast. The communicate part is easy to assume and hard to deliver. Deterministic networking means every sensor and actuator gets its data on a guaranteed schedule, not best effort, so a camera frame or a motor command arrives inside a bounded window every time, not just most of the time. That is the layer Ethernovia builds. Our HSB-enabled networking board for NVIDIA Holoscan platforms turns multi camera sensor data into deterministic, high bandwidth Ethernet feeding compute built for exactly that kind of coordinated system. Where does your stack's timing guarantee actually get tested? ethernovia.com #PhysicalAI #DeterministicNetworking #AutomotiveEthernet #Robotics

Ethernovia is currently building the underlying data backbone for the next wave of Physical AI for autonomous machines. In this role you will: -Own digital SoC design end to end: micro-architecture, RTL, verification, synthesis, lint, CDC, LEC and static timing closure -Work directly with system architects, software, hardware and verification engineers to take advanced automotive communication silicon from concept to tapeout -Help define an ASIC's interfaces, data flow and processing algorithms across multiple disciplines -Operate as a trusted self-starter who needs little guidance or oversight What we're looking for: 10+ years of ASIC RTL design or architecture experience, a proven track record shipping complex SoCs and strong command of Verilog/SystemVerilog. Background in Ethernet MAC/PHY, switching, SerDes integration or DSP filters is a plus. Hybrid in San Jose, CA. Pre-IPO equity, competitive salary ($200K-$300K) and a seat at the table while we scale from 100+ people to the next stage. Apply: https://lnkd.in/evu3pcqh #Hiring #ASICDesign #SemiconductorJobs #PhysicalAI #AutomotiveEthernet #Ethernovia

Einride and NVIDIA announced a new collaboration today adapting the NVIDIA DRIVE Hyperion platform for heavy duty freight trucks moving beyond controlled routes onto highways and suburban roads. Hyperion is billed as a production ready compute and sensor reference architecture. That phrase carries weight. An architecture proven on a passenger car does not automatically survive a Class 8 truck, where cable runs get longer, vibration gets harsher and sensor counts keep climbing as the safety case gets more demanding. Compute and AI models get most of the attention in announcements like this one. The physical layer carrying every camera frame and radar return from sensor to compute across that much steel and distance does not get a press release. It is what decides whether the reference architecture holds up outside a demo fleet. Scaling autonomy to thousands of trucks means scaling the network underneath it too. What part of your sensor stack would fail first at ten times the deployment scale? #AutonomousTrucking #PhysicalAI #AutomotiveEthernet #InVehicleNetworking

Goldman Sachs now projects millions of humanoid robots in operation by 2035, a number that reframes the robotics conversation from lab demos to actual supply chains. Getting there is not primarily a compute problem, it is a connectivity problem at industrial scale. Every humanoid robot packs dozens of joints, actuators and sensors into a frame smaller than a person, each one needing power and data with almost zero tolerance for latency or packet loss. That is the same deterministic networking challenge the automotive industry has spent two decades solving inside a vehicle chassis, just compressed into a smaller, faster moving package. Robotics companies scaling toward millions of units cannot afford to solve that problem from scratch. Automotive Ethernet standards and silicon built for functional safety, hardware level security and microsecond level timing are a proven starting point, not a lab experiment. Physical AI does not scale on compute alone. It scales on the nervous system underneath it. Where is your team seeing the biggest connectivity bottleneck as robot volumes climb? #PhysicalAI #Robotics #AutomotiveEthernet #TSN

Wire harness manufacturers are rethinking their production lines this month. The reason is architectural, not just electrical. A run of recent industry analysis makes the same point: zonal architecture is reshaping how the harness itself gets designed and assembled, not only what runs through it. Point to point wiring assumes every sensor and actuator gets its own dedicated run back to a central ECU. That means hundreds of individual wires, each one a manual routing step on the line. A zonal architecture replaces most of that with a handful of Ethernet links between zone controllers, which means a harness that is simpler to route, lighter to ship and easier to automate on the factory floor. That is the real prize hiding inside zonal architecture. Fewer ECUs get the headlines, but a harness a robot can actually assemble is what changes the economics of the vehicle. Ethernovia builds the automotive Ethernet backbone that makes that consolidation possible: deterministic, high bandwidth links that carry safety critical and infotainment traffic on the same physical wire without compromise. Where is harness complexity still slowing your production line down? ethernovia.com #ZonalArchitecture #AutomotiveEthernet #InVehicleNetworking #WireHarness

Every TSN schedule depends on a fact most people never think about: every node on the network has to agree on exactly what time it is, down to fractions of a microsecond. That agreement comes from IEEE 802.1AS, generalized Precision Time Protocol, gPTP. One device is elected grandmaster clock and every other node continuously corrects its own clock against it, accounting for cable delay and switch processing time at each hop. Without that shared clock, a scheduled window for a safety critical frame is just a guess, because every node's idea of "now" would drift apart within milliseconds. This is the part of automotive Ethernet that gets almost no attention because it has no dashboard number to point to. Bandwidth is easy to market. A synchronized clock across every zone gateway and endpoint is invisible until it fails. Every timing guarantee built on top of it fails right along with it. TSN scheduling gets the spotlight. gPTP is what makes the schedule mean anything at all. Where does your team spend more validation effort, the schedule itself or the clock underneath it? #TSN #AutomotiveEthernet #InVehicleNetworking #gPTP

A new market report values the automotive ECU market at 113 billion dollars this year, growing to 160 billion by 2033. The interesting part is not the total, it is what is driving it. Ford Motor Company's Universal EV platform cuts ECU count from roughly 70 in the F-150 Lightning down to just 5, running on a zonal architecture built on higher speed Ethernet. General Motors has gone further, developing a centralized computing platform for 2028 that connects propulsion, steering, braking, safety and infotainment through one Ethernet backbone, across gas and electric vehicles alike. These are not roadmap slides. They are production commitments from two of the largest automakers, both choosing Ethernet to tie braking to steering to infotainment, a different bar than winning a single ECU design. Consolidating dozens of controllers onto one backbone raises the stakes on every millisecond. A fault should not reach braking through a shared spine that also carries entertainment. Ethernovia builds automotive Ethernet solutions engineered for exactly that bar: deterministic timing and hardware level isolation so safety critical and infotainment traffic can share a backbone without sharing risk. Which of your vehicle domains would be hardest to trust on a shared backbone? ethernovia.com #ZonalArchitecture #AutomotiveEthernet #InVehicleNetworking #SoftwareDefinedVehicle

Every sensor, actuator and zone module in a modern vehicle or robot needs two things: power and data. For years those arrived through two separate connectors, two separate wiring runs and two separate points of failure. Hybrid connectors carrying both power and single pair Ethernet through one interface are changing that math. Instead of running a data cable and a power cable to every door module, camera or joint controller, one connector and one cable handle both. Fewer connectors means fewer crimp points, less harness mass and fewer places for a wiring fault to hide. This matters more as zonal architectures consolidate dozens of endpoints onto a handful of zone gateways. The wiring savings only materialize if the data side keeps its timing guarantees once it shares a connector with a power rail. Noise coupling and shared return paths do not disappear just because the interface got simpler. Where is harness complexity costing your team the most right now, connector count or cable routing? #ZonalArchitecture #AutomotiveEthernet #InVehicleNetworking #HarnessDesign

Renesas Electronics opened a Physical AI and Robotics Lab in Beijing this week. The numbers behind it are the interesting part. Today the company says it can address roughly 30 percent of a humanoid robot's bill of materials. The plan is to push that to 70 percent through control, power, sensing, AI and software. Notice what's missing from that list: the network tying all of it together. A humanoid robot is not one chip doing everything, it is dozens of sensors, actuators and compute nodes exchanging data within a hard latency budget, joint by joint, at kilohertz rates. As chip vendors race to own more of the bill of materials, someone still has to move the data between the pieces they build without adding jitter. That is the layer Ethernovia's HSB is built for, deterministic Ethernet as the nervous system connecting sensing, compute and actuation in real time. As physical AI hardware consolidates, is connectivity getting the same attention as compute and power? #PhysicalAI #Robotics #AutomotiveEthernet #InVehicleNetworking