China Mobile and Multiple Automakers Jointly Release Automotive Industry AI Foundation

On October 23, at the 2026 World Intelligent Connected Vehicles Conference, the National Artificial Intelligence Application Pilot Base (automotive direction in the manufacturing field, hereinafter referred to as the "Automotive Base"), led by China Mobile, officially released the first automotive industry AI foundation and its construction achievements to the public. Guests from more than 10 organizations, including the Beijing Economic-Technological Development Area Administrative Committee, Tsinghua University, FAW Group, Dongfeng Motor, Changan Automobile, BAIC Motor, and Li Auto, jointly witnessed the release ceremony. Adhering to the public service positioning of "neutral and not building cars, fully open capabilities," the Automotive Base's AI foundation capabilities, developed over the course of a year, are open for sharing across the entire industry starting today.

The automotive industry AI foundation delivers three major capabilities—computing power, data, and models—to the entire industry in an integrated manner, allowing automakers to use them on demand without building their own. The foundation currently manages 112.7 EFLOPS of computing power, with capabilities covering four major fields: R&D, manufacturing, intelligent driving, and testing. It has created six intelligent agents: automotive styling design, process planning, visual quality inspection, intelligent operations and maintenance, intelligent driving data generation, and intelligent driving model training. Related achievements have already been implemented at more than 10 central and state-owned automakers.

"Previously, automakers building their own AI systems generally faced the pain points of high investment and long cycles. The emergence of the AI foundation is equivalent to equipping the entire industry with the three core foundations of computing power, data, and models all at once. Automakers can reduce repeated investment in underlying infrastructure and focus on product and core technology R&D," said a responsible person from an automaker attending the conference at the release site.

Insufficient computing power, data silos, and high barriers to model development are the "three major difficulties" that have long constrained automakers' intelligent transformation—leading automakers face large investments and long cycles when building their own systems, while small and medium-sized automakers find it even harder to afford the huge costs. The automotive industry AI foundation released this time precisely targets these common pain points, packaging the three core capabilities of intelligent computing services, data services, and model services into public service capabilities: at the computing power level, automakers no longer need to spend heavily on building their own computing centers and can flexibly call the Automotive Base's public computing resources according to business needs; at the data level, automakers fully retain ownership of their core data, which circulates securely with ecosystem partners only under a compliance framework; at the model level, the Automotive Base provides a full-process toolchain from development and validation to deployment, lowering the threshold for implementing AI applications in the industry. Relying on the three major capabilities of the AI foundation, the large-scale application of AI in the automotive industry has been effectively promoted.

Six intelligent agents help automakers improve efficiency and upgrade in four major fields

In Shanghai, a design engineer at an automaker has firsthand experience of this: by handing sketches to the styling design intelligent agent, multimodal generation directly tripled the design cycle efficiency for a single set of design drawings; at another automaker in the same city, the process planning intelligent agent was deployed to the assembly workshop, and a process card for a single vehicle model that used to take five days to produce can now be completed in four hours.

In Jilin, the visual quality inspection intelligent agent raised the defect detection rate to over 99.5%, and line adaptation was compressed from four weeks to one week. The company's responsible person remarked that in the past, model changeovers were always held up by the adaptation cycle, and the application of the AI foundation genuinely shortened the overall business cycle; the automaker also applied the intelligent driving data generation intelligent agent developed by the Automotive Base, reducing intelligent driving R&D investment by 10% and significantly lowering the cost of real-vehicle road testing.

In Chongqing, automakers care more about production "not stopping": the intelligent operations and maintenance intelligent agent can identify more than 20 types of faults and risks, reducing fault downtime by 30%, easing concerns about continuous production.

In Beijing, a mobility company used the Automotive Base's intelligent driving model training intelligent agent to push the training acceleration ratio for autonomous driving strategies to 2x.

A responsible person from China Mobile said that in the next stage, the Automotive Base will continue to unleash the value of public services, open core capabilities, precisely align with automakers' scenario demands, promote technology pilots and application validation, refine and form a batch of replicable and promotable demonstration benchmark achievements, and work with upstream and downstream partners to jointly build and improve the automotive AI industry innovation ecosystem.