Mobileye's core ADAS business is the foundation: EyeQ chips powering L0-L2 safety features in over 230 million cumulative vehicles across 50+ OEMs. This segment generates nearly all of the company's revenue and all of its free cash flow. The business has the highest adjusted gross margin among auto semiconductor peers, yet trades at a steep discount to every major comparable.
Growth is the central tension. Revenue is guided flat to modestly up, weighed down by China share losses and EyeQ ASP erosion. Yet the pipeline is expanding as EU GSR2 mandates create a regulatory demand floor and Surround ADAS opens a higher-ASP product tier. The question is whether volume growth and new product categories can offset the competitive headwinds.
Why the discount may be temporary
If the multiple depression is primarily an Intel overhang effect (temporary governance and supply overhang), then any resolution of Intel's ownership stake could re-rate the core business toward peer multiples. The spread between current multiple and comp median implies significant upside if the overhang lifts.
What is the real switching cost for OEMs leaving Mobileye's integrated stack for Qualcomm/NVIDIA open platforms?
Mobileye's chip architecture is purpose-built for vision processing: highly efficient hardware accelerators that achieve state-of-the-art computer vision performance at a fraction of the power budget of general-purpose GPUs. The EyeQ6 family spans mass-market (EyeQ6L, single camera) to multi-sensor fusion (EyeQ6H, up to 11 sensors for Surround ADAS at 130 km/h hands-free driving).
Two emerging capabilities extend the technology moat. ACI (Artificial Community Intelligence) leverages the REM crowdsourced map from 230M+ vehicles for simulation-based training at enormous scale. VLSA (Vision-Language-Semantic-Action) adds a slow-thinking layer for scene understanding that goes beyond reactive driving. The open question is whether these efficiency-focused architectures can compete as end-to-end neural networks demand ever more raw compute.
Mobileye's OEM roster is the broadest in ADAS semiconductors: over 50 automakers use EyeQ chips. In 2025, the company won follow-on ADAS deals with every one of its top-10 customers and added two new OEMs. The depth of these relationships provides switching cost protection, as OEMs have years of integration work invested in Mobileye's software stack.
BMW precedent: open platforms win some OEMs
BMW phased out Mobileye for Qualcomm's integrated cockpit+ADAS platform (Neue Klasse). This loss demonstrates that some OEMs prefer open platforms where they control the software layer. The question is whether BMW is the leading edge of a broader shift or an outlier driven by BMW's specific in-house software ambitions.
China is the world's largest auto market and the fastest adopter of ADAS technology, but Mobileye is losing ground rapidly. Horizon Robotics dominates with nearly half the market, while NVIDIA surged in the NEV segment. Government procurement policies now give domestic suppliers a meaningful pricing advantage, accelerating the decoupling.
China may be structurally lost
The combination of domestic chip subsidies, procurement preferences, and Horizon Robotics's rapid share gain suggests China's ADAS market is structurally shifting away from foreign suppliers. Mobileye's remaining anchor in China is Zeekr (SuperVision), but even that relationship is reportedly under review. FAW/Hongqi is a partial offset but unlikely to reverse the trajectory.
Three competitors threaten Mobileye's ADAS dominance from different angles. Qualcomm wins with integrated cockpit+ADAS on a single chip, reducing OEM integration costs. NVIDIA wins on raw compute for premium autonomous systems, with the Uber partnership signaling L4 ambitions. Horizon Robotics dominates China through domestic advantages and government support. The common thread: open platforms where OEMs control their own software, versus Mobileye's integrated approach.
Black box vs. open platform: decisive or manageable?
Mobileye's integrated approach gives OEMs a turnkey ADAS solution but limits their software control. Qualcomm and NVIDIA offer open platforms where OEMs write their own stack. The question is whether this architectural preference is decisive for the majority of OEMs or primarily relevant to the handful (BMW, Tesla) that insist on full software control.