News from April 2025. Infineon agreed to buy Marvell's Automotive Ethernet business for $2.5 billion in cash, a significant milestone for automotive and semiconductor technology. The deal adds Marvell's Brightlane Ethernet PHYs, switches and bridges to Infineon's automotive microcontrollers and power devices. With the industry moving towards software-defined vehicles and far more data inside the car, the move positions Infineon to supply much of the in-vehicle network as well as the chips at its edges.
Key takeaways
- Price: $2.5 billion in cash, announced on April 7, 2025.
- Scope: Marvell's Brightlane automotive Ethernet portfolio of PHY transceivers, switches and bridges, with its team.
- Size: the business was projected to bring in $225 to $250 million of revenue in calendar 2025, at a gross margin of around 60%.
- Why: zonal, software-defined vehicles need high-bandwidth, low-latency networks, and Ethernet is replacing several older in-car buses for that job.
- Status: the deal closed on August 14, 2025 (see "What happened next" below).
What Infineon bought
Infineon plans to integrate Marvell's Brightlane portfolio, including high-speed Ethernet PHY transceivers, switches and bridges. A PHY is the chip that turns digital data into signals on the cable and back. A switch forwards traffic between many links. A bridge connects Ethernet to other interfaces inside an electronic control unit. Together they form the network that joins the computers in a car.
Infineon is already a leading supplier of automotive semiconductors, with its AURIX microcontrollers and a broad power portfolio. Pairing controllers with the network chips that connect them lets it sell a fuller zonal architecture: a central computer, several zone controllers and the Ethernet links between them. This adds scalable, low-latency communication to its offer for autonomous systems, advanced driver assistance systems (ADAS) and next-generation infotainment.
Why cars are moving to Ethernet
A modern car can carry well over a hundred electronic control units. Traditionally these talked over CAN, LIN, FlexRay and MOST, each with its own wiring. Those buses are reliable but slow by today's standards, and the wiring harness has become one of the heaviest and most expensive parts of the vehicle.
Cameras, radar and lidar for ADAS, high-resolution displays and over-the-air updates all push data rates far beyond what CAN can carry. Ethernet offers a single, well-understood protocol that scales from 10 Mb/s to multi-gigabit links, runs on a single twisted pair in automotive form, and can carry time-sensitive traffic using the IEEE Time-Sensitive Networking (TSN) standards.
Software-defined vehicles rely heavily on real-time communication and high data bandwidth, so they benefit directly from Ethernet's adaptability. In a zonal design, zone controllers gather local sensors and actuators and send their data over an Ethernet backbone to a central computer, where most of the vehicle's software runs. This acquisition aligns with the industry's shift towards more connected and more automated vehicles.
What to watch
- Integration: how quickly Brightlane PHYs and switches appear alongside AURIX controllers in reference designs.
- Competition: other suppliers of automotive Ethernet, including NXP, Broadcom, Texas Instruments and Microchip, are also active in this market.
- Standards: adoption of 10BASE-T1S at the edge and multi-gig links in the backbone will shape demand.
FAQ
How much did Infineon pay?
$2.5 billion in cash, as announced in April 2025.
What is Brightlane?
Marvell's brand for its automotive Ethernet products: PHY transceivers, switches and bridges designed for in-vehicle networks.
Why not keep using CAN?
CAN remains useful for simple control, but its data rate is far too low for cameras, sensor fusion and central computing. Ethernet scales from 10 Mb/s to 10 Gb/s and beyond with one protocol family.
Deep dive: the numbers and details
This part holds the data rates, a worked bandwidth example and the deal figures. Skip it if the overview is all you need.
In-vehicle networks compared
| In-vehicle network | Typical data rate | Common use |
|---|---|---|
| LIN | Up to about 20 kb/s | Seats, mirrors, simple switches |
| Classic CAN | Up to 1 Mb/s | Powertrain and body control |
| CAN FD | Several Mb/s in the data phase | Newer control networks |
| FlexRay | 10 Mb/s | Chassis and safety systems |
| 10BASE-T1S (IEEE 802.3cg) | 10 Mb/s, multidrop | Replacing CAN at the edge of zonal networks |
| 100BASE-T1 (IEEE 802.3bw) | 100 Mb/s | Cameras, diagnostics, infotainment |
| 1000BASE-T1 (IEEE 802.3bp) | 1 Gb/s | Zone-to-central backbone |
| Multi-gig T1 (IEEE 802.3ch) | 2.5, 5 and 10 Gb/s | High-bandwidth backbone and sensor links |
A worked example: camera bandwidth
Illustrative numbers. One uncompressed 2-megapixel camera at 30 frames per second with 12 bits per pixel produces about 2,000,000 x 12 x 30 = 720 Mb/s. Eight such cameras need nearly 6 Gb/s before radar, lidar and displays are counted. A 1 Mb/s CAN bus cannot move even one frame per second of that stream, which is why high-bandwidth links (Ethernet or dedicated camera SerDes) take over for sensor data while CAN stays for simple control.
The same arithmetic shows why the backbone needs multi-gigabit links. A single 1000BASE-T1 link carries one such camera with some headroom; a 10 Gb/s 802.3ch link could carry the eight-camera load above, with about 4 Gb/s left for everything else.
The deal in figures
| Item | Figure | Source |
|---|---|---|
| Price | $2.5 billion in cash | Marvell and Infineon, April 7, 2025 |
| Expected revenue | $225 to $250 million (Marvell: fiscal 2026; Infineon: calendar 2025) | Marvell release; Infineon closing release |
| Gross margin | Around 60% | Infineon |
| Design-win pipeline | About $4 billion by 2030 | Infineon closing release, August 2025 |
| Financing | Existing liquidity plus new debt | Infineon closing release |
The revenue is small relative to Infineon's total revenue, so the price, about ten times expected annual sales, reflects the expected growth of automotive Ethernet rather than current sales.
Global footprint
Infineon plans to use Marvell's technology and team to expand its US footprint and its research and development capability, strengthening its position in key global markets.
Broader applications: IoT and AI
While the main focus is automotive, the acquisition also opens doors in IoT and what Infineon calls physical AI. Ethernet's high bandwidth and low latency suit interconnected systems in robotics, smart factories and similar machines.
What happened next
On August 14, 2025 Infineon announced that it had completed the acquisition after receiving all regulatory approvals. It placed the business in a new "Ethernet Solutions" business line within its Automotive division, and cited a design-win pipeline of about $4 billion by 2030 (Infineon closing announcement, via Automotive World).
Takeaways
Infineon's acquisition of Marvell's Automotive Ethernet business is a significant step for automotive connectivity. By investing in Ethernet, Infineon is addressing the immediate needs of modern vehicles and laying groundwork for future uses in IoT and AI.
Reference links
- Marvell to Sell Automotive Ethernet Business to Infineon for $2.5 Billion in Cash (Marvell press release, April 7, 2025)
- Infineon successfully completes acquisition of Marvell's Automotive Ethernet business (August 14, 2025)
- Counterpoint Research: Infineon acquires Marvell's automotive Ethernet business to serve next-generation E/E architectures




Your comments will be moderated before it appears here.