ADI’s SDV: Not About the Computer, But Deconstructing the Architecture

ZONAL-READY wBMS

CES 2026’s SDV narrative often looks like a race for faster computers - bigger compute, more AI, flashier demos. But what Analog Devices, Inc.(ADI) presented was the opposite. They didn’t talk about “a stronger brain.” Instead, they demonstrated the structure of a nervous system - one that carries the brain’s commands to the furthest edge of the vehicle without loss, remains continuously verifiable, and stays updateable at all times.
Inside ADI’s private room demos, familiar domains - lighting, video, audio, and batteries - converged into one question: SDV is not ultimately about features, but about deconstructing the structure and redesigning it.

By | Sang Min Han _ han@autoelectronics.co.kr
한글로보기






CES 2026. ADI’s automotive demos are not on the show floor - they’re inside a private room. There are no concept cars designed to steal attention, no futuristic UI meant to perform “the future of user experience” on a stage. The approach is the exact opposite of that spectacle, yet within this restrained setting lies a sharper question. Because ADI never saw SDV as simply “the problem of putting a high-performance computer into a car.”
While semiconductor giants like NVIDIA, Qualcomm, and NXP concentrate on building smarter and more powerful “brains,” ADI is obsessed with ensuring that the brain’s commands reach every corner of the vehicle without degradation - while remaining diagnosable, provable, and updateable in real time. Yes, SDV is undeniably moving toward a centralized computing era. But no matter how powerful the center becomes, the moment that truly matters always happens at the edge. If sensors and actuators, wiring, power delivery, communications, diagnostics, and services cannot be bound into a single architecture, the “brain” becomes nothing more than an isolated device.
In that sense, ADI’s room is not just a showcase of technologies. It is a space that repeatedly reminds the industry of what SDV transition tends to forget. Where does control truly live? And how far should intelligence move to the center - and where must it remain at the edge? ADI answered not with features, but with structure.



Vision to Action
Control Logic Leaves the ECU and Flows Toward the Center

The booth tour began with lighting. It was a demo where a camera recognizes a person’s hand gestures and translates that output into headlamp behavior. On the surface, it looks like a simple and intuitive showcase: open and close your palm and the headlamps turn on; move your hand and brightness changes. But what ADI emphasized on site was not the novelty of the gesture - it was where the control logic lives.
ADI explained the demo as a structure combining GMSL (Gigabit Multimedia Serial Link) with E²B (Ethernet to the Edge). High-definition camera input is delivered via GMSL’s high-speed link into a central computing unit, where AI algorithms interpret hand motion, then send commands back down to control lighting behavior. The point is not that “the lights move,” but that the sense - decide - act loop is no longer scattered across edge ECUs - it is aligned into one centralized software flow. This is SDV’s promise of “software-defined functionality” expressed in its most concrete form.
On site, ADI framed E²B through an RCP (Remote Control Protocol) lens. It connects edge sensors and actuators over Ethernet while reducing the microcontroller (MCU) burden at the edge - pulling control and update responsibility upward into the zone or the central compute. In the headlamp demo, the message was repeated: there is no MCU at the edge, and software lives in the central computer.
This shift is not merely a change in wiring or communications. It also disrupts organizations and development workflows. When functions are scattered at the edge, each ECU has its own update path and test regime. But when functionality consolidates into the center, software moves as a larger unit and system integration accelerates - while verification and diagnostics become far more critical.



Centralized lighting control



SDV-Ready Headlamps
Edge ECUs Disappear, Software Remains (No MCU Required)

This view becomes even more explicit in the “No MCU Required” SDV-ready headlamp demo. Traditionally, headlamps were controlled by a closed structure - an ECU with its own intelligence (MCU) making local decisions. In an SDV transition, however, the headlamp no longer needs to be “a component that decides for itself.” ADI’s approach places direct lighting control under a zonal controller or centralized compute via Ethernet, while the edge becomes a simple execution layer.
Using E²B and LED drivers to drive a lamp matrix, the control logic runs in zonal-level software. Actuators become edge nodes, and the definition of functions moves upward into centralized software. This is not merely “a different control scheme for one component.” It is SDV’s core logic made tangible - the act of shifting functionality away from fixed hardware and into software.
The reason this structure is realistic is updates. On site, ADI explained the difference through the lens of OTA. When functionality is distributed across edge controllers, updating requires pushing software all the way to multiple ECUs and verifying each state. But when functions are defined centrally, updates become a “single central change.” The edge becomes a runtime node, and the update path simplifies. When SDV is described as “the evolution toward updateable cars,” the biggest transformation is not the feature itself - it is how update pathways and responsibility are restructured.



SDV-Ready Headlamps 



Centralization Doesn’t Remove Complexity

ADI does not describe SDV as an optimistic story only. As centralization progresses, systems do not simply become easier - they become complex in a different way. “Component-level complexity” may decrease, but “system integration and verification complexity” increases. At this point, what ADI calls the vehicle’s “nervous system” is not just about moving data quickly. It is a structure that reveals where the pain is the moment something breaks - and narrows down the root cause inside the system itself.
When SDV becomes real, the most frightening cost is not performance - it is an error that cannot be reproduced. ADI clearly understands that this complexity consumes time and money most brutally in development and production phases. That is why they strongly emphasized the GMSL Diagnostics demo.
Shown in an integrated environment featuring 12MP quad-camera surround view and 4K displays, the demo captures a crucial SDV reality: as ADAS and IVI become deeply integrated, communication devices operate under tighter timing and harsher link conditions. The risk is not only whether a feature works, but whether the integrated system remains stable in real conditions.
ADI’s question on site was simple: When black screens, flickers, or frozen images occur in a vehicle environment, how do we detect them with certainty? When a customer says “the display went dark once,” engineering teams spend endless time trying to reproduce the same symptom. And the fact that reproduction is difficult becomes even more lethal in the SDV era. The more complex the system becomes, the more “unreproducible issues” turn into costs that can break an organization.
That is why the demo centered on one thing: how data is left behind. ADI explained that every link status and event is collected at the SoC level, and each part’s condition is recorded with timestamps.



Centralization doesn’t remove complexity. The remaining battle is integration and verification. GMSL Diagnostics catches integrated-system issues not through reproduction, but through recording.



The Problem Is Reproduction Itself
Collect Everything with Timestamps

The demonstration was strikingly realistic. A cable was intentionally disconnected - the link dropped, and the screen went down. On the tool display, the fault location immediately appeared through color indicators and structural diagrams. As the view drilled deeper, a full system block diagram surfaced: cameras remained healthy, while a display link fault narrowed the source. Another step further: the DisplayPort input looked fine, but the link cable itself was broken - pinpointing the issue precisely.
The significance of this approach is not just “debugging becomes easier.” The biggest waste of time during development is attaching measurement equipment, repeating tests, and scraping logs endlessly. ADI emphasized that customers no longer need to connect oscilloscopes, multimeters, or I²C cables. If the system records states automatically and can be read instantly on site, SDV transition’s largest risk - development and operational cost - can be reduced meaningfully.
This diagnostics demo reframes SDV’s reality: centralization is not simplification by itself. What enables simplification is a structure built for diagnosability.



The Vehicle Is Not a Computer - It Becomes a Data Infrastructure

What makes ADI’s demos compelling is how they expand SDV beyond control and into an infrastructure of data, power, and experience. A demo focused on video and power delivery via a zonal SDV architecture is the clearest example. By supporting both USB-PD charging and DisplayPort Alternate Mode through USB Type-C, and distributing data via a combined pathway of GMSL, USB-C, and DisplayPort, ADI frames the vehicle as a platform-like infrastructure.
The most visually impressive moment is the seamless output of an uncompressed video stream. But the core is not video quality - it is the path. Once video data and power move together as one structure, infotainment becomes more than a screen. It becomes a service running on the vehicle’s network. In other words, SDV architecture begins to support not only control structure, but also the foundations of data flow, power distribution, and user experience.


 

A²B 2.0
In-Cabin Also Becomes Part of SDV

This direction continues inside infotainment. A²B 2.0 brings the in-vehicle audio network into SDV’s data flow rather than treating it as a standalone emotional layer. On site, A²B was introduced as “an audio bus that has already been used in cars for a long time.” It daisy-chains speakers and multiple audio nodes from the head unit, delivering both data and power through a single UTP (unshielded twisted pair) cable.
The key change in A²B 2.0 is bandwidth. On site, ADI emphasized a roughly 4x bandwidth increase versus the first generation, and the demo showed not only expanded audio streaming but also the ability to implement Ethernet tunneling over the A²B bus. In practice, a camera was connected to move data through the network while an FM audio stream played simultaneously and could be changed live. The message is clear: in-cabin audio is not a peripheral SDV function - it becomes a data flow residing on the same network structure.
One number becomes particularly powerful here. A²B 2.0 shows audio scalability not as “improved,” but as “structurally changed.” Audio may look like the language of experience, but under SDV it is redefined as a matter of data, updates, and network operations. A²B 2.0 makes it unmistakable that in-cabin is no longer a side feature - it becomes a node on the same backbone structure as SDV Ethernet.
That is why the statement - “A²B 2.0 can handle up to 119 channels of full-duplex audio upstream and downstream” - is not merely a boast. If SDV’s direction is not just larger computers but also the ability to move more data through thinner structures, in a more verifiable way, then in-cabin moves closer to SDV’s center. The competition is not “flashy UI,” but the infrastructure that holds it up. This is where ADI’s line - SDV is the work of deconstructing and redesigning structure - becomes sharper even in audio.



A²B 2.0: THE FUTURE OF INFOTAINMENT demo



wBMS
SDV Has Already Entered Production

The wireless EV battery management solution demonstrated that SDV is not a future concept. ADI’s wBMS (Wireless Battery Management System) entered production as early as 2021, and on site it was clearly confirmed as “in production with GM.” The starting point was not “wireless is convenient,” but manufacturability - simplifying processes and reducing costs from a production perspective.
On site, ADI presented a live-running module and explained that while it currently monitors two cells, it can be scaled to monitor many cells simultaneously in the same way. Then one more structural point was added: if wBMS traditionally sent information to a domain controller, ADI showcased an integrated direction that creates an Ethernet connection back to a central ECU, allowing the battery system to be treated like a zone. This reduces microcontroller complexity and cost at the edge while “future-proofing” the architecture - so that adding battery intelligence later does not explode software development burden.
In a domain like batteries, where safety and reliability are mandatory, a structure already in production proves SDV is no longer a lab concept. The architecture hardens first - software expands on top of it.



Not a Feature Change, But a Structural Transition

The strongest impression from ADI’s booth tour is less about “what was shown” and more about “what question was confirmed.” Across batteries (wBMS), video (GMSL), lighting (E²B), and audio (A²B 2.0), each demo delivers the same message. SDV is not the problem of installing a bigger computer - it is the problem of redesigning where control lives, how intelligence is distributed, and how wiring, power, and data pathways are structured. That is why this booth seems to talk about a “nervous system” rather than a “brain.”
No matter how strong the center becomes, what gives it meaning is the structure that runs through the entire vehicle. Competition in the SDV era will not end with “how much compute is installed.” It will be decided by how reliably that performance is delivered through the vehicle architecture, how quickly it can be verified, and how effortlessly it can be updated. ADI wanted to show exactly that.


   

---

Glossary
GMSL (Gigabit Multimedia Serial Link): ADI’s high-speed serial interface for transmitting high-resolution camera and display data
E²B (Ethernet to the Edge): A solution extending Ethernet to the edge using 10BASE-T1S (reducing edge MCU burden and strengthening centralized control)
A²B (Automotive Audio Bus): An automotive audio bus delivering audio data and power simultaneously over a single UTP cable; A²B 2.0 expands bandwidth and capabilities
wBMS (Wireless Battery Management System): ADI’s wireless battery management system enabling cell monitoring, with known production deployments

AEM(오토모티브일렉트로닉스매거진)

---