What is Automotive Ethernet? Complete Guide from CAN to 25GBASE-T1

What is Automotive Ethernet?

Automotive Ethernet is a vehicle-grade physical-layer protocol family that runs Ethernet over a single unshielded twisted pair (UTP) instead of the two or four pairs used in office Ethernet. The single-pair design cuts cable weight noticeably, halves connector pin count, and survives the −40 °C to +125 °C (−40 °F to +257 °F) temperature swings found in engine bays.

It is not a port of office IT Ethernet into vehicles. The IEEE 802.3 working group developed multiple new physical layers — 100BASE-T1 in 2015, 1000BASE-T1 in 2016, MultiGBASE-T1 (2.5/5/10G) in 2020, and 25GBASE-T1 in 2023 — specifically to meet automotive electromagnetic compatibility (EMC), latency, and reliability requirements. A separate low-speed multidrop variant, 10BASE-T1S, addresses the segments that traditionally belonged to CAN and LIN.

From the OEM perspective, Automotive Ethernet is the backbone for ADAS sensors, infotainment, gateway controllers, and increasingly powertrain telemetry. Industry projections (figures vary by analyst and program tier) anticipate a typical premium vehicle will contain dozens of Ethernet links by 2027–2028 — a roughly 10× increase over 2020 levels, driven mainly by ADAS sensor fusion and zonal architectures.

Why CAN and SerDes Are Not Enough for High-Bandwidth Links

Automotive Ethernet does not replace CAN bus, LIN, or FlexRay. It complements them. Three protocols dominated vehicle networks for decades and continue to ship in volume: CAN (1986) for body and powertrain control at up to 1 Mbps (CAN FD up to 8 Mbps), LIN for low-cost local subnets, and SerDes families (FPD-Link III, GMSL) for camera and display video. Each remains the right choice for its segment.

What changed is that ADAS and infotainment created a new bandwidth tier neither CAN nor SerDes can serve as a network. A 1080p/30fps front camera can generate hundreds of Mbps to more than 1 Gbps of payload, depending on pixel format, color depth and any in-sensor compression or preprocessing. A LiDAR sensor outputs 100–400 Mbps. A modern infotainment head unit serving rear-seat displays can saturate a 1 Gbps link. CAN cannot carry any of this; SerDes can carry one stream per link but is point-to-point, not switchable like Ethernet.

Three structural advantages drove the move to Ethernet for high-bandwidth links:

• Switchable network topology. Ethernet supports multiple devices behind a single switch, with VLANs, QoS and TSN. SerDes is point-to-point; CAN is a multi-master bus, but speed-limited.
• Wiring weight. Ethernet's single-pair construction replaces multiple parallel buses for high-bandwidth links and produces a noticeable harness weight reduction in zonal architectures (industry estimates published by Bosch and Continental whitepapers fall in the high-single-digit kilogram range per vehicle).
• Cost per node at scale. Each SerDes link needs a dedicated SerDes chip pair per camera. Ethernet uses standard PHYs that are shared across applications and can run common AUTOSAR / SOME-IP / DoIP stacks on top.

Most 2024–2030 vehicles will run a hybrid architecture: CAN FD and LIN for legacy body and low-speed control, FlexRay where it remains qualified, SerDes for high-resolution display and camera links, and Ethernet for the high-bandwidth backbone, ADAS sensor fusion and central compute. 10BASE-T1S is the bridge that lets Ethernet eventually cover the segments CAN occupies today.

Automotive Ethernet Standards: 10BASE-T1S to 25GBASE-T1

Five electrical IEEE standards cover the bandwidth tiers shipping today, plus one optical specification for high-speed and long-distance links. Each runs over a single twisted pair (or, in the optical case, glass fiber) but uses different modulation, cable specifications and topology.

For 100BASE-T1 and 1000BASE-T1, an unshielded twisted pair (UTP) with 100 Ω characteristic impedance is sufficient for most cabin applications. MultiGBASE-T1 (2.5 Gb/s and above) typically requires shielded twisted pair (STP) or shielded differential pair (SDP) to meet the OPEN Alliance EMC limits, especially in engine bay or near-motor locations. 25GBASE-T1 nearly always uses SDP construction.

Connector Families: Differential T1 vs Coax / Quad-Coax / SerDes

The connector landscape splits into two physical-layer groups, and confusing them is one of the most common RFQ mistakes we see. Group 1 is true single-pair differential connectors designed for IEEE 802.3 T1 PHYs. Group 2 is coax, quad-coax, and SerDes-oriented connectors that often ship next to Automotive Ethernet links but are not native single-pair T1 hardware.

Group 1: True Differential T1 / Automotive Ethernet Connectors

Group 2: Coax / Quad-Coax / SerDes / Legacy High-Speed Vehicle Connectors

These connectors originated for RF antenna feeds and SerDes video (FPD-Link III, GMSL). They are still widely deployed and often appear next to Automotive Ethernet links, but they are not native single-pair T1 hardware. HSD belongs in this group: it is a star-quad connector originally designed for LVDS camera links, not for T1 differential Ethernet PHYs.

The market is consolidating. Most analyst forecasts expect H-MTD, the multi-gig differential families (GEMnet, NETBridge), and Mini FAKRA / HFM for SerDes camera links to dominate new designs through 2030. FAKRA and HSD remain in legacy production but are declining in new high-speed program awards.

For deeper pinout and selection details on each family:

• FAKRA Connector Pinout and Color Codes
• Mini FAKRA / HFM Selection Guide
• HSD Connector Pinout and Star Quad Wiring
• H-MTD Specifications for Multi-Gig Automotive Ethernet
• MATEnet Modular Architecture

OPEN Alliance and IEEE 802.3 Compliance

Two standards bodies govern Automotive Ethernet. IEEE 802.3 defines the physical and data-link layers. The OPEN Alliance — an industry consortium of OEMs and suppliers — defines the test specifications that cables, connectors, and ECUs must pass to qualify for vehicle production.

The OPEN Alliance Technical Committees (TCs) most cable and connector suppliers must understand:

• TC1 — 100BASE-T1 Channel. Original channel and link segment specifications for 100 Mbps cables.
• TC8 — ECU / PHY Test Specification. Compliance and interoperability test for 100BASE-T1 / 1000BASE-T1 ECUs and PHY implementations. Cable, connector and link-segment specifications are covered separately under TC9.
• TC9 — Channel & Components. Component-level specifications for the link segment: cable, connectors, in-line connections. Covers 100/1000 BASE-T1 channel components.
• TC10 — Sleep / Wake-Up. Power management and sleep/wake protocol specification (not multi-gig cable, despite occasional confusion).
• TC15 — Multi-Gig Interoperability and Compliance Tests. Test specifications for 2.5/5/10GBASE-T1 interoperability and compliance.

For an OEM to add a cable supplier to its approved vendor list, the supplier typically must submit:

• Test reports from a TÜV-, DEKRA- or UL-accredited lab covering the relevant OPEN Alliance TC specifications
• An IATF 16949 quality management certificate
• Material declarations meeting REACH and IMDS requirements
• PPAP / APQP documentation aligned with the OEM's program

Typical Applications: ADAS, Camera, Infotainment, Powertrain

Automotive Ethernet now spans every domain in the vehicle. The bandwidth required varies by application; the connector and cable choice follows from there. The table below maps current production mappings — it will shift toward higher bandwidth tiers as L3+ ADAS and zonal architectures become mainstream.

* Front-camera payload depends on pixel format, color depth, and any in-sensor compression. Raw uncompressed video can exceed 1 Gbps; compressed streams are typically lower. Always confirm with the camera datasheet.

For deeper application breakdowns, see our ADAS Camera Cable Selection Guide and Robotaxi Cable Solutions for Self-Driving.

Beyond the physical layer, Automotive Ethernet relies on a software stack — TSN (Time-Sensitive Networking, IEEE 802.1), AUTOSAR, SOME/IP for service-oriented communication, and DoIP (Diagnostics over IP) for service. These are covered in our forthcoming Automotive Networking Software Stack guide.

RFQ Checklist: How to Specify an Automotive Ethernet Cable

Cable selection comes down to the same ten fields on every OEM and Tier-1 RFQ we receive. Filling these in upfront cuts quotation time from weeks to days and prevents requoting after qualification surprises.

Common RFQ patterns we see (China-side OEM and Tier-2 perspective)

From supplier-side RFQ review for China-based automotive cable programs, two patterns recur in European, North American, and Asian RFQs:

1. "Cat5e for Automotive Ethernet" — First-time buyers often ask for Cat5e because it sounds equivalent to 100BASE-TX. It is not. The cable will pass an in-house bench test and fail OEM TC9 channel verification later in the qualification cycle, costing weeks of program slip.
2. FAKRA for new 1000BASE-T1 designs — Migrating from FAKRA to Mini FAKRA / HFM or to a true differential connector (H-MTD, MATEnet) is now the industry default for new ADAS programs targeting 1 Gbps and above. RFQs that specify FAKRA for 1G links typically get a follow-up call asking whether the program can move to a differential connector before tooling is committed.

For a complete RFQ template you can send directly to suppliers, see our Custom Automotive Ethernet Cable RFQ Guide.

Future Outlook: 25 Gbps Today, 50 Gbps and Beyond

The frontier has already moved. IEEE 802.3cy-2023 standardized 25GBASE-T1 for 25 Gb/s on a single balanced pair. The remaining challenge is not standardization itself — it is production adoption: cable and connector qualification, EMC margin at multi-GHz frequencies, and PHY power consumption at 25 Gbps. Most Tier-1 suppliers expect 25GBASE-T1 in production vehicles by 2027–2028.

For 50 Gb/s automotive links, the standardized path today is optical. IEEE 802.3cz-2023 defines 2.5 / 5 / 10 / 25 / 50 Gb/s over glass fiber for in-vehicle and longer-reach automotive applications. There is currently no published IEEE standard for electrical single-pair 50GBASE-T1; suppliers that mention "50G electrical Automotive Ethernet" are referring to roadmap, not a ratified spec.

Practical implications for cable suppliers and OEMs preparing programs that launch in 2027–2030:

• Specify connectors with a documented upgrade path. H-MTD and the multi-gig differential families have published roadmaps to 25 Gbps and beyond. FAKRA does not.
• Plan cable construction for the worst-case channel. SDP construction qualified at 10GBASE-T1 has a better chance of passing 25GBASE-T1 channel tests than a cable barely qualified at 5G.
• For 50 Gb/s and routes that cross EMC-hostile zones, evaluate optical (802.3cz) seriously — it eliminates EMC at the cost of a more complex termination.