Mercedes-Benz to develop self-driving cars with NVIDIA - Qazinform

Mercedes-Benz locking in NVIDIA is not a procurement decision—it is a strategic capitulation of European automotive to the CUDA ecosystem, abandoning the fantasy of in-house silicon. Technical cascade: Automotive compute is irreversibly shifting from distributed ECU clusters to centralized architecture. NVIDIA's Thor platform (2000 TOPS) will directly displace Bosch and Continental domain controllers. Once sensor-fusion algorithms are compiled into the CUDA toolchain and OTA-validated in production, switching costs escalate exponentially—mirroring the ARM-to-custom-silicon trap that locked smartphone OEMs in 2019. Supply-chain exposure: NVIDIA's advanced-node fabrication depends on TSMC in Taiwan, China. Every revision to the US export-control list is a disruption vector for European OEMs. More critically, outsourcing the full train-infer loop for driving algorithms to a US entity creates friction with the EU AI Act's data-sovereignty mandates for high-risk AI systems. Compliance costs will ultimately land on end consumers. Competitive dynamics: Tesla's vertical FSD integration and Huawei's chip-algorithm-cloud trinity make Mercedes' outsourced-compute choice a tacit admission that European AI talent density cannot match Silicon Valley. Mobileye and Qualcomm Snapdragon Ride will see order books compress further, eroding Tier-1 bargaining power. 12–24 month outlook: BMW and VW will likely follow within 18 months, pushing NVIDIA's automotive revenue growth past 80%. The true long-tail effect: once all three European majors run on CUDA, NVIDIA graduates from supplier to de-facto standard-setter, with pricing power far exceeding today's level.