After Nvidia CEO Jensen Huang stepped down, Chinese self‑driving companies are accelerating a Nvidia backup plan that relies on locally produced AI chips and efficiency measures.
Jensen Huang resignation shakes AI supply chain
Nvidia founder and chief executive Jensen Huang announced his resignation this week after admitting to an alleged affair with a suspected spy, a development that sent shockwaves through the global AI ecosystem. Huang had long championed open‑source initiatives in the United States, positioning Nvidia as the de‑facto hardware backbone for autonomous‑vehicle development. The abrupt leadership change raises immediate questions about the stability of Nvidia’s long‑term commitments to automotive partners, especially those in China that depend on its GPUs for model training and its automotive processors for on‑board inference. Analysts at Bloomberg note that any perceived uncertainty could accelerate existing discussions about alternative suppliers.
China’s self‑driving sector already faces a precarious balance between soaring demand and constrained supply of high‑performance chips. Companies such as Mercedes‑Benz, Zoox and Wayve have built full‑stack partnerships with Nvidia, yet many Chinese startups operate with a mix of Nvidia GPUs for training and third‑party processors for inference. The potential for stricter export controls, a scenario that gained traction after Huang’s exit, adds a layer of risk that could disrupt production pipelines and delay rollout schedules for autonomous fleets across major Chinese cities.

Pony.ai pivots to domestic chips as a backup
Pony.ai’s chief financial officer Leo Wang told TechCrunch Mobility that the firm is actively pursuing a two‑pronged strategy to mitigate chip shortages. First, engineers are being instructed to maximise efficiency by opting for mid‑tier compute resources rather than the most expensive high‑end GPUs, a move that reduces cost while preserving training throughput. Second, Pony.ai is increasingly integrating domestically produced AI chips, which have matured considerably over the past few years and now offer a viable, albeit slightly less cost‑effective, alternative to Nvidia’s offerings.
Efficiency measures
Wang explained that the shift toward mid‑tier hardware involves re‑optimising model architectures and adopting mixed‑precision training techniques, allowing the same workloads to run on less powerful accelerators. This approach not only eases pressure on Nvidia’s supply chain but also aligns with broader industry trends toward sustainable AI compute.
Domestic alternatives
China’s home‑grown chipmakers have accelerated development of AI accelerators tailored for autonomous‑driving workloads, and Pony.ai is already testing several of these silicon solutions in pilot projects. While Wang admitted that the cost‑per‑performance ratio still favours Nvidia, the growing ecosystem provides the company with a credible Plan B that could be scaled if export restrictions tighten further.
Broader industry moves amid export‑control concerns
Beyond Pony.ai, other Chinese and international players are adjusting their roadmaps to hedge against potential Nvidia supply disruptions. May Mobility, for example, announced a partnership with Volkswagen to deploy the all‑electric ID. Buzz as a base vehicle for its global ride‑hailing platform, a deal unveiled at Nvidia’s GTC event in Berlin. The arrangement signals that automakers are preparing modular platforms that can accommodate a variety of autonomous‑driving stacks, including those that may rely on non‑Nvidia hardware.
Funding activity also reflects the shifting landscape. Japanese robotaxi startup Newmo secured a ¥20.8 billion ($135 million) round led by KDDI, while Dutch parking‑tech firm ParkBee acquired Mobian’s city‑parking operations. European urban‑logistics company Port raised €14.5 million from the European Bank for Reconstruction and Development and Balnord. These capital infusions suggest investors are betting on diversified technology stacks and alternative supply chains to sustain growth in the autonomous‑mobility market.
Outlook for China’s autonomous‑vehicle sector
TechTimes assesses that the emergence of a Nvidia backup plan could reshape competitive dynamics in China’s autonomous‑vehicle arena. While Nvidia’s GPUs remain the performance benchmark, the maturation of domestic chips offers manufacturers a degree of strategic flexibility that may reduce reliance on U.S.‑origin technology. If export controls become more stringent, Chinese firms are likely to accelerate the adoption of locally sourced silicon, even if it entails higher short‑term costs.
In the longer term, the industry’s ability to balance cost, performance and regulatory risk will determine the pace of autonomous‑vehicle deployments across Chinese megacities. Observers expect that the next wave of vehicle‑to‑cloud services will incorporate hybrid compute architectures, blending Nvidia’s high‑end solutions with emerging Chinese accelerators, thereby creating a more resilient ecosystem for the country’s ambitious self‑driving ambitions.
TechTimes Perspective
TechTimes believes the fallout from Jensen Huang’s resignation highlights the fragility of a supply chain that has become overly dependent on a single U.S. vendor. For Chinese autonomous‑vehicle firms, developing a robust Nvidia backup plan with domestic chips may not only safeguard against export restrictions but also stimulate local semiconductor innovation. The move could narrow the performance gap over time, though price differentials may persist. Vietnamese operators and investors should watch for price signals on imported Nvidia hardware, as any tariff or licensing shift could affect the cost structure of future autonomous‑mobility services in the region.







