<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/38b4b4e2-491a-4aae-82b6-2ec370c742a4.png?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/38b4b4e2-491a-4aae-82b6-2ec370c742a4.png"/>
Author | Wang Xiaojuan
Editor | Zhou Zhiyu
On March 23, the Chinese autonomous driving industry welcomed two major pieces of news on the same day.
On this day, QCraft announced the completion of its Series D funding round, raising $100 million. The investor lineup was quite notable, including a leading domestic vehicle manufacturer, a leading automotive electronics component company, and several industrial funds.
This company, which started with L4 autonomous driving and later established a foothold in the L2++ mass production market, has clearly directed its funds toward R&amp;D in the directions of world models and reinforcement learning. Its CEO, Yu Qian, said, &#34;Autonomous driving is the best gateway to physical AI.&#34;
On the same day, XPeng announced the establishment of its Robotaxi business unit.
According to Wallstreetcn, this primary organization of XPeng will oversee the product definition, project integration, R&amp;D testing, and operations of Robotaxi, with Yuan Tingting, Senior Director of Product at XPeng's Autonomous Driving Center, serving as the head. Empowered by the second-generation VLA (Vision-Language-Action) large model, XPeng plans to commence passenger-carrying demonstration operations in the second half of this year and launch three Robotaxi models in 2026.
At the recent NVIDIA GTC conference and various industry summits, CEO Jensen Huang repeatedly stated: &#34;The next wave of artificial intelligence is physical AI. AI will understand the laws of the physical world, and autonomous vehicles are the largest and most mature embodied intelligent robots we can currently see.&#34;
It's not just QCraft and XPeng.
In early March, WeRide deepened its cooperation with Geely Farizon, planning to deliver 2,000 factory-installed mass-produced Robotaxis by 2026. Pony.ai joined forces with Toyota and GAC Toyota, aiming to deploy a fleet of thousands of vehicles in first-tier Chinese cities. GAC OnTime completed the delivery of hundreds of vehicles, doubling its fleet size to 600. CAOCAO INC has set up over 3,600 virtual pickup and drop-off points for Robotaxi in Hangzhou.
After two years of silence, the autonomous driving track collectively sprang to life in the spring of 2026.
<h2>Three Bottlenecks Loosened Simultaneously</h2>
Two years ago, L4 autonomous driving was a completely different picture.
Around 2024, this track was still about burning money and telling stories. The hardware-stacking solution composed of lidar, high-compute chips, and high-precision maps made the cost of retrofitting a single vehicle easily reach hundreds of thousands of yuan. Robotaxi demonstration operations in certain cities were more like money-burning shows, far from true commercialization.
The capital market's patience for technology stories ran out, pushing a batch of L4 startups to the brink of survival. The core logic of the industry was forced to shift from &#34;running faster&#34; to &#34;living longer.&#34;
An autonomous driving industry insider recently told Wallstreetcn that since 2025, changes in technology, hardware costs, and policies have pushed the autonomous driving industry to a new node.
First, the technological paths have converged.
End-to-end large models have become an industry consensus, and the L4 technical route has been redefined. The Tesla CyberCab officially rolled off the production line in February 2026, proving that a pure vision solution and end-to-end large models can support vehicles operating for extended periods and distances without human intervention. XPeng's second-generation VLA model also achieved end-to-end direct generation from visual signals to action commands.
More importantly, NVIDIA released Alpamayo at CES, the world's first open-source autonomous driving VLA model with thinking and reasoning capabilities, while simultaneously opening the high-fidelity simulation framework AlpaSim and large-scale driving datasets. This built a three-in-one open ecosystem of &#34;model-simulation-data,&#34; significantly lowering the R&amp;D threshold for advanced autonomous driving.
Second, the establishment of the factory-installed mass production route.
Compared to earlier retrofitted Robotaxis, the industry is now generally shifting to the factory-installed mass production route. WeRide's GXR is equipped with the latest autonomous driving suite GEN8 and relies on Geely Farizon's steer-by-wire AI chassis, supply chain, and production control system. The production cycle time for a single vehicle has been significantly reduced from 1 hour to under 10 minutes.
Pony.ai's seventh-generation Robotaxi autonomous driving suite has seen its total cost reduced by 70% compared to the previous generation, with the in-vehicle computing unit cost decreasing by 80% and lidar costs by 68%. The selected vehicle models now cost between 100,000 and 150,000 yuan.
With costs reduced to this level, the single-vehicle profitability model for Robotaxi can finally become viable.
Third, a substantial breakthrough in policy bottlenecks.
In December 2025, the Ministry of Industry and Information Technology announced the first batch of L3 autonomous driving vehicle access permits. BAIC ARCFOX Alpha-S and Changan Deepal SL03 models subsequently conducted road tests in designated areas in Beijing and Chongqing. As of mid-January 2026, Deepal's L3 vehicles had accumulated over 70,000 kilometers of autonomous driving.
More importantly, the pilot program clarified the primary responsibility of vehicle manufacturers during system activation, resolving the long-standing issue of liability allocation in the industry. Although L3 and L4 still differ legally, this signal has opened the door for the commercialization of L4 autonomous driving technology.
Ouyang Minggao, an academician of the Chinese Academy of Sciences, predicted at the 2026 High-Level Forum on the Development of Intelligent Electric Vehicles media release that by 2030, L4 autonomous driving based on advanced end-to-end large models will achieve large-scale commercialization in mid-to-high-end passenger vehicles.
<h2>Physical AI Is No Longer Just a PPT</h2>
&#34;In 2026, humanity is at a critical watershed moment in AI development. The next 5 to 10 years will see greater opportunities in the physical world.&#34; QCraft CEO Yu Qian closely linked the development of autonomous driving with the advancement of AI while announcing the financing.
Physical AI has become the industry's hottest new story.
This concept is attractive because it elevates autonomous driving from a vertical track to a universal gateway for AI to enter the physical world.
At its technology day last year, XPeng upgraded its brand positioning to &#34;Explorer of the Physical AI World,&#34; with its VLA model capable of driving four major carriers: automobiles, Robotaxis, humanoid robots, and flying cars. NVIDIA's Alpamayo model, released at CES, also promoted the banner of physical AI.
But what's different in 2026 is that the narrative is accelerating into quantifiable business metrics.
From a technical perspective, the computing power race has not disappeared but has entered a new phase. XPeng's Robotaxi plans to be equipped with 4 Turing AI chips, offering vehicle-side computing power of up to 3000 TOPS, and adopts a pure vision solution that does not rely on lidar or high-precision maps.
NVIDIA introduced the Alpamayo platform, attempting to package the &#34;brain + skull&#34; for automakers to lower the deployment threshold for advanced autonomous driving. Hard indicators for evaluating algorithm capabilities now include a 50% reduction in end-to-end latency, a 20% increase in traffic efficiency, and a 30% reduction in harsh braking rates.
Mass production scale is also expanding.
WeRide plans to deliver 2,000 factory-installed mass-produced Robotaxis in 2026, and Pony.ai aims to expand its fleet to 3,000 vehicles. QCraft's &#34;Chengfeng&#34; intelligent driving-assist system has been installed in over 1 million vehicles, with cooperation from nearly 10 leading automakers. The company expects to add over 50 new cooperative models in 2026.
The prerequisite for scale is controllable costs. Some institutions predict that if annual production reaches 100,000 units, the manufacturing cost per Robotaxi could drop to $10,000.
From a commercial operations perspective, Robotaxi is starting to make financial sense.
Waymo's paid rides have reached 450,000 per week, and its operating range has expanded to international markets such as Houston, Miami, Tokyo, and London. Domestically, QCraft has partnered with the industrial chain to officially enter the unmanned logistics vehicle sector, operating in Jinhua, Wuhu, Ningbo, and other places, pioneering the &#34;mass production is operation&#34; paradigm. It is estimated that the market size of Robotaxis in China will grow to $8.655 billion by 2033, with a compound annual growth rate of 74.0% from 2025 to 2033.
Deutsche Bank, in a report after CES, assessed: &#34;2026 will be the year autonomous vehicles transition from testing/verification to scaled deployment, and humanoid robots will move from laboratory experiments to small-scale deployment.&#34;
This recovery is not just technically driven but is a synchronized evolution of technology, policy, cost, and user perception.
Users are no longer obsessed with ever-expanding technical specifications but pursue intelligent driving functions that are easy to use, reliable, and affordable. The penetration rate of urban NOA reached 15.1% from January to November 2025 and is accelerating toward models priced below 200,000 yuan. The L3 pilot accumulated over 70,000 kilometers of autonomous driving in 19 days, covering complex urban scenarios such as overpasses and congested roads.
Of course, the industry is not without its concerns. Some industry insiders told Wallstreetcn that uncertainties remain in the pace of policy and legislation, systemic safety incidents could trigger regulatory tightening, and it remains unknown whether cost reduction speeds can keep up with the intensity of price wars.
The survival space for third-party intelligent driving solution providers is being squeezed by the full-stack in-house R&amp;D of leading automakers, and the threshold of the &#34;technical moat&#34; is rising rather than falling.
In 2026, the autonomous driving industry no longer relies on stories for financing but begins to speak with numbers. Physical AI has driven from the exhibition booths onto city streets, and L3 has transitioned from policy documents into daily commutes. After a decade-long marathon, the inflection point may have truly arrived.
Market risk exists; investment requires caution. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are appropriate for their specific circumstances. Investing based on this information is at your own risk.

02643

00800

XPEV

09868

DRIV

On March 23, the Chinese autonomous driving industry welcomed major news: QCraft completed a $100 million Series D funding round, with funds directed toward the research and development of world models and reinforcement learning; XPeng established its Robotaxi business unit, planning to launch three Robotaxi models in 2026. Multiple companies in the industry, such as WeRide and CAOCAO INC, are also accelerating their layouts, marking the recovery of the autonomous driving sector

- On March 23, 2026, the autonomous driving industry in China experienced significant developments, including Lightyear Technology's $100 million funding round and XPeng's establishment of a Robotaxi division.  
- These advancements are part of a broader transition towards L4 autonomous driving, with improvements in technology, lower costs, and favorable policy developments prompting a renewed interest in the market.  
- Analysts predict that the Robotaxi market could reach $8.655 billion in China by 2033, reflecting a compound annual growth rate of 74.0% from 2025 to 2033.

Has the Inflection Point for Autonomous Driving Truly Arrived?