Gasgoo Munich-At electronica Shanghai 2026, STMicroelectronics (ST) showcased a rather unique trunk demo. Two zone controllers, linked via an Ethernet ring network, drove speakers in their respective zones. The sound remained synchronized with precision imperceptible to the human ear. There was no independent external amplifier box, no A2B bus—just a single Ethernet backbone running through it.
ST insists this isn't merely a proof of concept; several local OEMs and Tier 1 suppliers have already secured the solution for POC (proof of concept) studies.
The core of this solution is the Stellar G6 MCU, which features a built-in Ethernet switch and audio interfaces. As vehicle E/E architectures shift toward central domain control, audio systems are stepping out of their isolated amplifier boxes and plugging into the zone controller network.
Yet, we have to ask: Is this really necessary?
The Heavy Burden of Centralized Audio
To understand why distributed audio is becoming an option, we first need to look at the problems plaguing traditional setups.
A typical mid-to-high-end in-car audio system includes a standalone external amplifier box (usually tucked under a seat or in the trunk). Inside, amplifier chips are stacked to drive anywhere from a dozen to over twenty speakers. This box requires dedicated thermal management—metal casing paired with active or passive cooling. It’s bulky, heavy, and limits packaging flexibility.
But a more fundamental issue lies with the A2B bus. An audio-specific bus standard championed by ADI, A2B transmits audio data and control signals over a single twisted pair while powering peripherals like microphones. For the past decade, it has been the de facto standard for automotive audio.
However, A2B is essentially a daisy chain. All audio data ultimately converges on the amplifier box for centralized processing. This topology clashes with the current evolution of vehicle E/E architectures from distributed to zonal setups. Zone control logic divides functions by physical area, whereas A2B requires all audio data to travel on independent channels.
Then there’s the often-overlooked factor: weight. As EVs become increasingly sensitive to curb weight, A2B-specific wiring harnesses, the standalone amplifier box, and its thermal structure are all coming under scrutiny.
Moving from centralized to distributed, the audio system is treading a path already taken by the powertrain and ADAS domains: integrating independent ECUs into higher-level controllers.
ST's Edge: PCM, Built-in Switch, and the ARM Ecosystem
Distributed audio makes logical sense, but implementation requires hardware support. ST’s Stellar G6 positions itself along several key dimensions.
First, the unique status of PCM memory.
In the race for new process nodes in automotive MCUs, the choice of NVM (non-volatile memory) is a decision every manufacturer must face. ST chose PCM (phase-change memory), while many competitors opted for RRAM or MRAM. In terms of mass production progress, ST’s PCM is indeed ahead. The Stellar series has been in mass production globally for about two years, with shipments reaching millions of units. Zhang Rujing, senior marketing manager at STMicroelectronics, emphasized to Gasgoo: "It is currently the only new NVM process that has been automotive-qualified, shipped over a million units, and validated by OEM DV testing."
Zhang Rujing, Senior Marketing Manager at STMicroelectronics
For OEMs, the choice of memory comes down to supply chain certainty. Against the backdrop of persistent fluctuations in automotive chip capacity, a mass-production solution already validated by Design Verification (DV) is far more persuasive than one still in the sampling stage. ST’s PCM density retains an advantage at the 28-nanometer node, allowing the Stellar G6 to achieve roughly 30MB of storage capacity—sufficient for the OTA and functional integration needs of zone controllers.
Second, the system cost narrative of the built-in Ethernet switch.
The Stellar G6 integrates a TSN-standard Ethernet switch internally, eliminating the need for external switch chips and their surrounding power circuits. Joint calculations by ST and a leading OEM suggest the entire distributed audio system can save tens of dollars per vehicle. This includes the amplifier box, A2B transceivers, wiring harnesses, and connectors.
Tens of dollars on a BOM (Bill of Materials) list is no small sum. With price competition in the auto market yet to abate, any convincing system-level cost reduction will draw OEM attention. It is worth noting, however, that these calculations are based on a specific model; savings will vary significantly across different vehicles.
Third, the ecosystem advantages of the ARM R52+ and hardware Hypervisor.
ST’s shift from PowerPC to ARM architecture was driven largely by ecosystem considerations. The ARM R52+ is an automotive functional safety-enhanced core co-defined by ST, Bosch, and ARM. It offers ready-to-use resources in development toolchains, software libraries, and talent pools.
The hardware Hypervisor built into the Stellar G6 deserves a mention. When applications with different safety levels—such as body QM and chassis ASIL D—need to be deployed on the same MCU, the Hypervisor provides independent isolation. It even allows different development teams to work in parallel on the same chip. Zhang revealed to Gasgoo that the AUTOSAR standard recently announced support for Hypervisor hardware virtualization, adding that "ST has already pre-implemented these features in advance based on these requirements."
Fourth, deep collaboration with partners.
In early 2026, ST and a renowned automotive infrastructure software company jointly released a ZCU distributed audio solution based on Ethernet. The company provides a full TSN protocol stack, including clock synchronization, audio transport, and traffic shaping. Combined with the Stellar G6’s hardware capabilities, they claim end-to-end audio latency of less than 2 milliseconds and jitter under 100 microseconds. If these metrics are verified in mass-produced vehicles, the solution could meet the stringent latency requirements of scenarios like active noise cancellation.
From Demo to Mass Production: Industry Challenges Remain
Although ST’s solution is technically robust, pushing distributed audio from a demo to mass production faces several critical hurdles.
The first is the restructuring of industry roles. In traditional audio systems, tuning, sound processing, and speaker matching are typically handled by specialized audio suppliers (such as Harman, Bose, or DSP Concepts). In a distributed setup, defining the boundary between a Tier 1’s hardware delivery responsibilities and an audio software provider’s algorithm responsibilities—and establishing an ASPICE-compliant development process—remains a gray area in the current division of labor.
The second is the latency requirements for active noise cancellation. Road noise cancellation requires a closed loop—from microphone pickup to speaker output of anti-noise waves—to be completed within milliseconds. While automotive Ethernet’s TSN protocol guarantees deterministic low latency at the transport layer, eliminating cumulative jitter caused by OS scheduling and cross-core data migration in a complex chain involving Stellar G6, central computing units, and multiple DSPs remains an engineering challenge. Zhang revealed that ST is jointly developing active noise cancellation solutions with third parties, though a specific mass production timeline is not yet clear.
The third is the industry’s path dependence on A2B. ADI’s A2B technology enjoys high penetration and a mature ecosystem in automotive audio. With the latest generation of A2B chips further enhancing flexibility and node capacity, Ethernet AVB/TSN must prove more than just bandwidth advantages to fully replace A2B. It also needs breakthroughs in single-node PHY cost, the simplicity of daisy-chain implementation, and cable costs.
Another variable worth watching is RISC-V. ST currently maintains a cautious wait-and-see attitude toward RISC-V, primarily concerned about software compatibility issues arising from ecosystem fragmentation and the difficulty of achieving high functional safety certification (ASIL-D). In contrast, some Chinese chipmakers are actively pushing RISC-V automotive MCUs. If RISC-V can establish a unified functional safety standard ecosystem within the next 3 to 5 years, ST’s commitment to the ARM route could face severe challenges in terms of cost-effectiveness and flexibility.
In Conclusion
Distributed audio is not a demand created out of thin air. The zonal evolution of vehicle E/E architectures is a genuine trend, and the audio system cannot remain an outlier forever. From this perspective, ST’s Stellar G6 solution is heading in the right direction.
But we must also recognize that audio systems have their own peculiarities. Their requirements for latency, jitter, and synchronization precision are far stricter than those of body control. Moreover, they involve a complete ecosystem spanning algorithms, tuning, and amplifier matching. There is still a significant amount of engineering work to be done between a demo and mass production, from POC to SOP (Start of Production).
As Zhang emphasized during our discussion: "ST is not a low-price company; we are a technology-driven company." In the fiercely competitive arena of automotive MCUs, being technology-driven implies longer product definition cycles and higher R&D investment. But it also means building genuine barriers once the product matures.
Will distributed audio become standard in the next generation of smart vehicles? There is no answer yet. But one fact is certain: when the core logic of E/E architecture shifts from "functional domains" to "zones," every system originally organized by domain—power, chassis, body, cockpit, audio—deserves to be re-examined.
Audio is just one link in this systemic reconstruction. And every change must ultimately answer the same question: Does it make the car lighter, cheaper, and better? If the answer is yes, it will eventually appear on the BOM list of a mass-produced car.
