Published on Oct. 21, 2026

At the Paris Automotive Summit on October 15th, 2026, XPeng launched XPENG AI Engineering, a new business that opens the company’s in-house smart mobility solution technology to automakers and partners around the world, the Chinese automobile portal Gasgoo reported.

The offer has three layers. Partners can license full vehicle platforms, including structural and electrical/electronic architectures built on 10-gigabit Ethernet. They can also adopt individual modules: the Turing AI driving system, powered by XPeng’s second-generation VLA model, the Turing AI chip, an AI BOX capable of running a 30-billion-parameter MoE model, and an 8C battery that charges from 10% to 80% in eight minutes.

The third layer is XPeng’s engineering toolchains for calibration, end-of-line testing and AI diagnostics. XPeng says these cut inspection time by 41 minutes per vehicle and annual electrical-testing costs on the production line by 95%.

He Xiaopeng pointed to XPeng’s partnership with Volkswagen as proof that the model works, and set out an ambitious roadmap. NGP, XPeng’s navigation-assisted driving, has begun its first large-scale public test in Europe, with customer rollout planned from 2027, and the IRON humanoid robot is due for mass production in late 2026. The move signals a broader shift: under pressure at home, Chinese EV makers are starting to sell their AI know-how, not just their cars.

XHEART, a chip designer for physical AI that is less than three years old, has closed a strategic funding round with a notable mix of backers. Renault Group invested for the first time, Chery added to its stake, and memory maker CXMT joined, a rare bet by CXMT on an automotive chip company, reported the finanical outlet STAR Market Daily (in Chinese).

XHEART’s cumulative funding now stands at about 5 billion yuan (around US$750 million). Its investors also include the National AI Industry Investment Fund and two National Social Security Fund vehicles.

The company is led by Li Jun, Momenta’s former head of chips. Its core product, the X7, targets urban assisted driving (L2++) and L3/L4 autonomy. The chip debuted in SAIC Volkswagen’s ID.ERA 9X and now ships in MG, Roewe, Cadillac and Chery models.

Industry insiders told the tech portal 36Kr that Momenta helped shape the X7 from the planning stage, so it is designed around how the driving algorithms actually run instead of carrying redundant capacity.

CXMT’s involvement is the strategic heart of the deal. Memory prices have soared this year, and supply now decides chip contracts. One insider told 36Kr that Qualcomm won a Leapmotor order partly because an earlier deal with Micron gave it cheaper memory. CXMT already mass-produces automotive-grade LPDDR5X, so it could give XHEART a similar edge on price and supply, the report said.

State-owned CHN Energy said on October 15 that its Power Marketing Company and its Research Institute (the National Institute of Clean-and-Low-Carbon Energy) have built an AI power trading platform, the group’s newspaper, Guojia Nengyuan Jituan Bao reported.

The platform combines models of how power markets operate, including their trading rules, with data-driven AI. It runs across the whole trading workflow, from price forecasting and strategy generation to bid submission and post-trade review. It is designed for traders and the AI to work together.

Its nodal price forecasts were 3–6% more accurate than mainstream methods. In electricity retail, after deviation penalties, the AI bought power more cheaply than human traders on 59.5% of trading days. In coal-fired generation, after operating and start-up compensation, the AI strategy earned a cumulative return 39% higher than the human plan, the company claimed.

As China builds a unified national power market during the 15th Five-Year Plan period (2026–2030), spot trading is expanding, rules are getting more complex and price swings are widening. For generators, AI-assisted trading is fast becoming a core capability rather than an experiment.

At this year’s China International Industry Fair (CIIF) in Shanghai, Black Lake Technologies gave its first centralized showcase of decision-making AI and made clear who it sees as its core market, reported 21st Century Business Herald (in Chinese).

Co-founder Xiao Zhe expects some of the strongest demand to come from small and medium-sized manufacturers. They are eager to adopt and willing to pay, he said, as long as the AI solves real problems on the shop floor.

Black Lake’s answer is a suite of more than ten industrial agents covering engineering, quoting, scheduling, order tracking, quality inspection and warehousing. These are the everyday functions that smaller factories often run with thin teams.

The engineering agent, for example, can analyze drawings and process requirements in about one minute with 95% accuracy, after which engineers review and adjust the results. Xiao stressed that, unlike general-purpose AI, an industrial agent has to understand a factory’s business rules, resources and real-time status, and key production decisions still need a human sign-off.

According to Shanghai Daily, Black Lake’s platform is already used in 30,000 factories. Client Mixue Ice Cream & Tea linked orders, raw materials, production and quality control through the system. It reports feature development cycles more than 70% shorter and overall efficiency up 30%.