Gasgoo Munich-The previous two installments, "Why is RISC-V Suddenly Hot?" and "How Does RISC-V Overcome Fragmentation?", clarified two points. They explained why RISC-V is suddenly trending and how its fragmentation risks might be mitigated. But the industry is more concerned with practical questions. Can this free architecture actually be mass-produced in vehicles? Where will it appear first? Who is paying for it? Ultimately, how will it reshape the automotive chip landscape?
Hype aside, the automotive market ultimately rewards only hard cash and mass production. This article skips concepts and standards to track its real progress in the industry.
Starting with Control Chips
A new chip architecture rarely enters a vehicle at the core right away, and RISC-V is no exception.
One production scenario already running is the headlight controller. Zijing Semiconductor, incubated by Great Wall Motor's technical center, has deployed its first RISC-V automotive chip, the M100. It is used in body control scenarios like combination headlights, ambient lighting, and switches—up to 17 chips per vehicle. According to Zijing, the M100 entered mass production by the end of 2025. It is reportedly the first RISC-V automotive chip to do so globally. Great Wall has also outlined a plan to equip 2.5 million vehicles within five years.
Beyond body control, powertrain is another landing zone. Dongfeng’s RISC-V multi-core engine control chip, the DF30, has completed validation on multiple models. Mass production is slated for 2026.
Image Source: Dongfeng Motor
Industry insiders reveal that since August 2024, several Chinese RISC-V automotive MCUs have hit mass production, with another batch set to arrive in 2026. Applications are concentrated in control and specialized scenarios like body, powertrain, and electronic rearview mirrors. The CPU cores in these control chips are entirely RISC-V, serving as the main controllers.
For larger chips in autonomous driving and smart cockpits, it currently appears mostly as an auxiliary core or a dedicated acceleration unit. The Horizon Robotics Journey 6 series has partially adopted RISC-V cores and entered mass production. However, its main control CPU remains ARM. Mobileye’s next-generation EyeQ Ultra autonomous driving chip pushes further. According to official data, all 12 general-purpose CPU cores on the chip are RISC-V. Yet, this remains a rare case in the high-end chip market.
Why are big chips so hard to replace? Ye Qi, senior manager at SAIC Motor's passenger vehicle architecture and systems department, notes a common mindset. Automakers often "prefer expensive foreign chips over the risk of switching." Domestic chipmakers can only gradually break in through edge scenarios like lights and wipers.
He points out that the high-end MCU market has long been dominated by NXP, Infineon, and Renesas. They hold roughly 70% of the share and are deeply intertwined with the AUTOSAR toolchain. Meanwhile, smart driving and cockpit chips are led by Qualcomm, MediaTek, and Nvidia, where software and algorithm barriers are equally high. Swapping a chip implies redoing the hardware, basic software, and testing processes from scratch.
The breakthrough in edge scenarios like lights and wipers is driven by risk considerations. These control chips are used in high volumes per vehicle and have relatively independent functions. Single-point failures don't lead to loss of vehicle control. The verification scope is manageable, and pin-compatible replacements are possible. This makes the cost of trial and error relatively controllable. Moreover, while low-end control cores widely use ARM, their ecosystem lock-in is mostly limited to toolchains and development habits. It is also the zone of fiercest price competition.
The industry has chosen an "edge-to-center" path. It first achieves mass production in body and power control chips. Then, it enters autonomous driving chips as general CPU cores, control cores, or specialized acceleration units. Finally, it gradually extends into cockpit control and central computing chips.
Changan has taken a middle road. After evaluating control cores, Changan did not target replacing main control chips in large systems. Instead, it leveraged RISC-V's custom instruction extensions to build a low-latency matrix computing acceleration prototype for on-vehicle AI inference. Ding Ke summarized the strategy as "pushing control chips down to the IP level." Automakers may not manufacture chips themselves, but they are starting to decide which cores go inside and what extensions to reserve.
The timing points in this direction too. Ye Qi estimates that 2026 to 2028 is the window for the shift toward electronic-electric architectures based on central computing plus regional control. The landscape for new product categories like regional controllers isn't fixed yet. The penetration rate of new architectures is expected to exceed 30% by 2027.
In his view, the rollout of L3 regulations, widespread adoption of City NOA, and volume expansion of 800V platforms will drive chip demand. This will occur during this same period. New categories have room to be redefined from the outset. An architecture generation switch means re-selection. Re-selection is the moment when entering the supply chain is cheapest for a new architecture.
The Costs Beyond "Free"
"Free" is the advantage most frequently cited for RISC-V. Yet, when automakers and chip companies run the numbers, they find other expenses inevitably pop up despite saving on licensing fees.
According to industry insiders, ARM core licensing fees range from several million to over 10 million yuan depending on the tier. Post-production royalties are calculated as a percentage of the chip price. In comparison, commercial RISC-V core licensing fees are roughly one-tenth of ARM's, with lower royalties. Public industry data shows ARM licensing fees generally sit in the million to ten-million dollar range. Royalties mostly range between 1% and 2% of the chip price, varying by agreement.
To be clear, only the instruction set itself is free. According to the RISC-V International website, RISC-V waives licensing and royalties at the instruction set level, but that doesn't mean the chips are free. Commercial core IP, development tools, and verification environments based on RISC-V still require payment. The core value for automakers adopting RISC-V lies not in obtaining "free chips." Rather, it breaks the lock-in of existing suppliers. This provides bargaining leverage and a strategic "second option."
The bigger expense lies in the ecosystem. Liu Lin, senior manager for Infineon's automotive business and head of RISC-V ecosystem in Greater China, laid out their timeline at the conference. The company began building the ecosystem two or three years before sample chips were available. From 2022 to 2024, they worked on compilers, debuggers, and virtual prototypes. From 2025 to 2026, they will fill in operating systems and basic software. Only after 2027 will application layer reference designs like motor drives and AI acceleration arrive. In March 2026, Infineon released the DRIVECORE, a pre-integrated evaluation package for RISC-V virtual prototypes. This allows customers and partners to start adaptation before hardware is ready.
Image Source: Infineon Website
Even the global leader in automotive MCUs has to invest in the ecosystem years before a chip launches, so the barrier for small and medium players is even higher. In his speech, Li Weili noted that a full automotive-grade chip typically takes three to five years to complete certifications like functional safety. Engineers who understand both functional safety and the new architecture, along with certified toolchains and safety software libraries, are currently scarce resources.
Ecosystem investment is too heavy for one company to shoulder alone, which is one reason the automotive RISC-V alliance Quintauris was formed. Founding members like Bosch, Infineon, NXP, and Qualcomm jointly maintain underlying software and automotive real-time profiles. The more manufacturers participate, the less each pays in duplicated learning costs. This also explains why the standards and compatibility mentioned in the previous article are so important. They essentially help the entire industry share the burden of repeated investment.
Digging deeper, the question becomes whether automakers should enter the fray and how. Yang Quanshen articulated the dilemma clearly at the conference. If an automaker develops chips in-house but production volumes don't scale, R&D costs won't be amortized, making iteration difficult. Yet, if they buy fully integrated "black box" solutions, they risk losing control over algorithms—essentially surrendering their core assets. He positioned Axera as a "pure computing enabler," sharing profits with OEMs while leaving technical room for Tier 1 suppliers.
RISC-V offers a middle ground: the architecture is open, allowing automakers to participate deeply in definitions without designing everything from scratch. Great Wall has taken this a step further by incubating Zijing Semiconductor within its technical center. The OEM directly states requirements, and the chip team designs to order. Zijing claims this model shortens the automotive chip development cycle by about 40%. Dongfeng’s DF30, meanwhile, took the route of tackling challenges jointly with Chinese IP and chip companies.
Ultimately, automakers care about three things: supply security, cost, and autonomy. These translate to: Will I be cut off? Is it too expensive? Who calls the shots in the future? The difference in licensing fees is the easiest part to calculate. Ecosystem investment and the division of labor are the heavyweights that determine whether RISC-V can succeed. Much of the money saved on licensing fees is redirected into commercial IP, toolchains, and the automakers' own teams.
Coexistence, Not Substitution
Whether RISC-V will replace ARM is the question every discussion eventually lands on. Yet, the answer from both inside and outside the conference hall is remarkably consistent. It is complementary in the short term and coexistence in the long term—not one replacing the other.
Zijing Semiconductor’s view is "complementary, not replacement." RISC-V opens up ground in MCUs and body control through cost and customization advantages, while ARM continues to lead in high-performance computing. Infineon’s product portfolio is a direct example: the next-generation AURIX MCU will switch to RISC-V, but the ARM-based TRAVEO series remains on sale. Qualcomm is a dominant force in ARM-based cockpit chips, yet it is also a founding member of the automotive RISC-V alliance Quintauris. The giants are hedging their bets, which speaks volumes about their judgment of the future.
More giants are choosing to enter first and see where the road leads. NXP is a founding member of Quintauris, and Renesas has launched RISC-V products in industrial sectors, though it remains cautious in automotive. In this review, neither has announced a mass production timeline for automotive RISC-V MCUs. For giants with solid market share, moving early means bearing the cost of an immature ecosystem, while moving late risks losing customers. It’s not surprising that they are securing their position in the ecosystem first, then deciding on product timing.
Image Source: Quintauris Website
However, the certainty of growth is not low. In his speech, Li Weili cited predictions showing RISC-V’s penetration in the automotive sector is expected to rise significantly. It will grow from about 10% between 2025 and 2026 to roughly 31% by 2031, based on data from organizations like RISC-V International. But even if this prediction materializes, nearly 70% of the market will still belong to other architectures—it’s hardly a total sweep.
In terms of rhythm, control chips will lead and large chips will follow—there is almost no suspense. This is because the closer you get to the center, the less the outcome depends on architecture licensing fees. It depends more on decades of accumulated software legacy, development ecosystems, and deployment data. These happen to be the weakest links for any new architecture.
China is driving this shift the fastest. Automakers are personally defining chips, with Great Wall, Dongfeng, and Changan each taking their own paths. The density of Chinese RISC-V automotive MCU mass production is rare globally. Yet, amidst the heat, there is a hidden concern. The greatest dividend of an open-source architecture is that the world shares a set of underlying blueprints and an ecosystem. If everyone closes their doors to build mutually incompatible extensions, it amounts to turning open source into a new form of closed source.
Liu Lin specifically noted at the conference that Chinese standards must align with international ones. Infineon participates in both Chinese working groups and Quintauris efforts, and Quintauris has already begun cooperating with Nuclei System Technology. Only if this path succeeds can China's mass production experience become part of the global ecosystem, rather than standing as an isolated island.
The window of opportunity Ye Qi described is also ticking down. In his assessment, the triple windows of architecture switching, demand explosion, and standard maturation are all concentrated between 2026 and 2030. Before that window closes, control chips must achieve mass installation. The software ecosystem must also cross a critical threshold, or RISC-V could remain stuck at the edge control layer. Conversely, once benchmark models and data from millions of installed units are proven, the speed of penetration toward the center could exceed current expectations.
Putting this information together, the contours of the future landscape are already fairly clear. For the next several years, automotive chips will enter a rare era of multi-architecture coexistence. In control layers like body and power, RISC-V’s share will rise rapidly. Main control CPUs for smart driving and cockpits will remain dominated by ARM. Cases like EyeQ Ultra will be hard to replicate in the short term. At the same time, RISC-V will continue to permeate large chips in the guise of auxiliary cores and specialized accelerators.
Most of these changes won't appear at press conferences, and consumers won't buy a car just because it uses RISC-V. However, increased supplier competition will ultimately translate into lower costs and stable supply. The real transformation happens at the negotiating table and on procurement lists. Once the choice of underlying architecture shifts from the hands of a single company to the entire industry, it is difficult to take back.
This trilogy boils down to three sentences: popularity is just the start, standards solve coordination, but ultimately, mass production has the final say. The true value RISC-V offers the automotive industry was never a free core. It is an additional pivot point for supply chain security and more leverage in defining the industry. This foundational change is measured in decades—it has only just begun.
