China's 2030 Autonomous Plan Just Killed the 'China Lag' Trade — Tesla, Mobileye, Pony AI, and WeRide Are Now on Beijing's Clock

On Thursday, September 11, 2026, China's Ministry of Industry and Information Technology (MIIT) published the roadmap for the 15th Five-Year Plan covering smart electric vehicles. Reuters and CNEVPost confirmed three numbers in lockstep: new-energy vehicles are to reach 70% of domestic new passenger-car sales and 40% of commercial-vehicle sales by 2030, and autonomous-driving vehicles are to see large-scale deployment on highways, urban expressways, and selected city roads by that same date. The ministry also signalled tighter rules on vehicle production, road safety, self-driving, and software upgrades — meaning the2030 target is enforceable, not aspirational.

The Society of Automotive Engineers' forecast that frames the policy is sharper than the MIIT language: by 2030, 20% of new cars sold in China will be fully driverless, and 70% will carry advanced assisted-driving features. That is a meaningful step beyond the 70% NEV share figure, because AV software is where most of the value migrates inside an electric car. For US-listed AV names, the question is no longer whether China catches up; the question is who sells into the transition.

Investors who shorted or underweighted Chinese AV names on the assumption that US autonomy leads were working from outdated premises. Baidu's Apollo Go completed 3.4 million fully driverless rides in Q1 2026 alone, with the platform now operating in 15 cities. Robotaxi analyst models already point to500,000 robotaxis on Chinese roads by 2030 — versus roughly 4,100 today — and a market that scales from $4.4 billion in 2024 to roughly $120 billion by 2030. The data flow, not the policy, is what killed the lag thesis.

Tesla finally rolled out Full Self-Driving (Supervised) in China in May 2026 after years of data-localization delays. The global FSD subscriber base reached 1.48 million by the end of Q2 2026, up 56% year on year, and the company delivered a record 480,126 vehicles in the quarter — its best Q2 ever. Those are the headline numbers; the China-specific ones are weaker.

Tesla's China-market EV share collapsed to 3.2% in October 2025 even as Chinese OEMs were busy deploying L2++-class ADAS — the same XPeng Highway Pilot and Li Auto AD Max systems that have already been trained on Chinese road data for years. The MIIT plan turns that head start into a structural advantage: by 2030, 70% of new Chinese passenger cars must carry advanced ADAS, and the standards now being enforced are written around the architectures Chinese OEMs and Mobileye jointly developed.

Mobileye sold $235 million of product into China in the first half of 2026 — 22% of its $1,066 million in H1 revenue, up from $201 million in the year-earlier half. The EyeQ SoC family that powers Chinese OEM ADAS accounts for roughly 91% of total revenue, and the share shipped to Chinese OEMs is rising. In other words, the US-listed ADAS champion is already the largest non-domestic supplier into the very market the MIIT plan is reshaping.

Mobileye raised its full-year 2026 revenue guidance midpoint by $20 million to roughly $1.995 billion and lifted its adjusted operating-income guidance by 88% after a one-time R&D incentive under Israeli law. That backdrop matters: the company is profitable on an adjusted basis in2026 for the first time in several quarters, and the MIIT roadmap guarantees that Chinese OEM ADAS demand will keep climbing into the 2030 deployment target. SuperVision — Mobileye's higher-margin level-3-capable system — is now on its second OEM design win, and the third Surround ADAS customer (Mahindra) signals the platform is broadening beyond its Geely-and-Polestar base.

Pony AI and WeRide are the two US-listed Chinese robotaxi operators, and the MIIT plan reshapes their market structure more than anyone else's. Pony AI sits at a $125M TTM revenue base with Q2 2026 quarterly revenue up 68.8% year on year, a16% gross margin, and a stated goal of deploying 100,000 autonomous trucks on Chinese roads by 2030. WeRide carries $830.8M in TTM revenue but is heavily loss-making, with an enterprise value of $7.93B against only $1.73B of equity market cap — a function of capital intensity rather than market scepticism.

The market structure argument is that Pony AI and WeRide are no longer racing each other for a small pilot pool — they are chasing a 500,000-fleet, $120B TAM with regulatory certainty. Baidu's Apollo Go, the dominant incumbent at 3.4 million Q1 2026 fully driverless rides and 15-city coverage, is the benchmark they have to out-execute on cost per mile and unit economics. Goldman Sachs's fleet forecast moves from 4,100 in 2025 to roughly 1.9 million by 2035 — implying2026–2030 alone absorbs ~500,000 vehicles, with the operator that controls the fleet-management stack capturing the bulk of recurring revenue.

Hesai is the cleanest beneficiary in the lidar complex. The company delivered RMB 860.8 million ($126.9 million) of Q2 2026 revenue — up 21.9% year on year — and shipped 628,275 lidar units in the quarter. ADAS lidar shipments grew 60% YoY to over 485,000 units and robotics lidar grew 193% to over 142,000 units. Hesai holds roughly 44% of China's automotive lidar market and 61% of global robotaxi lidar, and posted positive net income for the full year 2025 — the first Western-listed pure-play lidar company to do so.

Upstream of the perception layer sits the compute stack. MIIT's separate infrastructure plan targets 9,800 exaflops of national intelligent computing capacity by 2030 — more than quadruple the June 2026 baseline — backed by roughly3.8 trillion yuan ($532 billion) of cumulative information-infrastructure investment. NVIDIA is the structural supplier to that build-out, even as Chinese OEMs increasingly diversify into domestic AI accelerators (Huawei Ascend, Cambricon, etc.). The August 2026 GB 44721-2026 standard sets the technical floor for L3/L4 perception systems, and Hesai's product roadmap is built directly to that spec.