Momenta: Approaching Break-even

"The concept that autonomous driving naturally involves continuous burning of money is a misconception. Managing the profitability rhythm and operating cash flow well is the greatest responsibility to customers."

What gives the Momenta executives the confidence to say this is its first impressive performance report released after listing.

This performance data is stunning: For the six months ended June 30, 2026, the company achieved operating revenue of 1.6 billion RMB, a 75.9% year-on-year increase; gross profit of 1.17 billion RMB, a 79.4% year-on-year increase; adjusted net loss was only 14.1 million RMB, a significant contraction of 96.6% year-on-year.

It is worth noting that against the backdrop where the autonomous driving industry still generally faces profitability pressure, this data means Momenta's core operations are basically break-even.

In terms of sales, Momenta's production-grade intelligent driving solutions had new installations of approximately 321,000 units in the first half of the year, a year-on-year increase of 83.7%; cumulative production models reached 105, covering 26 OEM clients; cumulative designated models 219, an increase of 49 models compared to the end of 2025; 114 designated orders on hand yet to be delivered.

It can be seen that Momenta's "data flywheel" is already turning.

Regarding the route choice, Momenta CEO Cao Xudong said: "Many people compare the endpoint of autonomous driving to going to the moon, then feel Mount Everest is closest to the moon, so they start climbing it immediately. But we chose another path: first figure out what is needed to go to the moon - we need to build rockets. Production-grade autonomous driving is the rocket we are building; it provides data fuel; large-scale commercialization of L4 is the ultimate moon landing.""

Regarding the price war, company executives said: "We will absolutely not blindly follow the trend to slash prices to the bottom just to secure orders. Intelligent driving concerns life safety; cheap but unsafe systems will directly backfire on the car brand's reputation. "

Momenta also revealed that it is discussing subscription pricing models targeting consumers with OEMs. The industry case of Tesla Version 14 having a subscription rate of about 50% in the first half provides a reference for intelligent driving shifting from one-time licensing to recurring revenue. With the implementation of L3 and improved regulations, Momenta's business model is expected to upgrade from single-vehicle license to "hardware + software + subscription services".

Adopting the "overwhelming" strategy to achieve predictable growth

Against the background of the Chinese market where new car lifecycles are short and iteration speed is extremely fast, Momenta adopted the "overwhelming" strategy, meaning the company does not bet on a single model or a single customer, but covers as many OEMs and model matrices as possible, avoiding sales volatility risks through the law of large numbers.

This strategy has clearly been successful - multiple models equipped with Momenta's solution became hits. The new BMW iX3 had orders exceeding 30,000 in two days of launch; the all-new Mercedes-Benz GLC SUV EV had over 12,000 reservations within 24 hours of going online; some domestic top models can achieve a monthly sales volume of about 10,000 units.

On the other hand, Momenta's delivery efficiency is also significantly improving. From the first 100,000 units taking 24 months to install, to now requiring only 36 days for every additional 100,000 units, Momenta's mainstream platform and automated toolchain are transforming "delivery capability" from project-based to platform-based capabilities. Company management stated at the earnings call that their delivery efficiency has evolved from "400 people, one car in two years" to "dozens of people, one car in three months", behind which is the continuous construction of mainstream reuse, tooling, and automation system capabilities.

R7 World Model: "3-5x Performance Leap" of the Technical Foundation

The core highlight at the technical level is undoubtedly Momenta R7 World Model. This model has entered the delivery phase in the third quarter of 2026.

Cao Xudong detailed the three-layer architecture of the world model: Pre-training (relying on over 1 million vehicles, 13 billion miles of driving data), Simulation (10 million kilometers of simulated mileage run daily, equivalent to over 4 billion kilometers of real road testing), Reinforcement Learning (system autonomous evolution, surpassing human driving levels).

Compared to the previous generation R6 model, R7 has achieved a 3-5x performance improvement in multiple core scenarios, and does not require customers to change sensors and chips, only software upgrades are needed to achieve it. In the second half of the year, heavyweight models such as Volkswagen E29X, Cadillac X5, Mercedes-Benz GLC/GLE, BMW i3, etc., will sequentially be equipped with R7 for production.

The deeper meaning of the world model lies in enabling Momenta to use the same technical foundation to horizontally connect passenger cars, Robovan, Robotaxi, and even future embodied intelligence scenarios. This "one foundation, multi-scenario reuse" architecture not only dilutes R&D costs but also allows real data generated in different scenarios to feed back into model iteration.

At the company level, it is also emphasizing that its "Four-Dimensional Flywheel" is accelerating: Data Dimension - Total driving data will reach 24 billion kilometers by the end of 2026; Computing Power Dimension - GPU expanded from 20,000 to 60,000 units. This "Data-Computing Power-Model-Business" self-cycling system is the core logic for Momenta to build long-term barriers.

Momenta's Choice, "Building Rockets Rather than Climbing Mount Everest"

From Cao Xudong's perspective, the autonomous driving industry has had two completely different approaches from the beginning.

As mentioned earlier, Momenta did not choose the path of most people, but started with the end in mind, "first think clearly what the essence of going to the moon is, what is needed to solve this essential problem - need to build rockets. What we chose is to down-to-earth build rockets."

This path was not clever in the early stage. When others tried to run dazzling demos with a small-scale fleet in a limited area, Momenta chose a slower, heavier, and harder-to-replicate path: first put L2+ production vehicles on the road on a large scale, accumulate massive real driving data, drive algorithm iteration with the data flywheel, and finally feed back to the leap of L4 capabilities.

Cao Xudong summarized this as the company's "Dual-Path Strategy": one leg is production autonomous driving, the other leg is fully autonomous driving. Production autonomous driving is responsible for providing data fuel - equivalent to building rockets; L4 large-scale commercialization landing is the ultimate "moon landing" goal.

Cao Xudong gave a market judgment, After 5 years, the global Robotaxi market capacity is expected to reach 1 to 1.5 million units, Robovan market capacity about 3 million units, the combined total exceeds 20 billion USD."We are targeting the truly scalable Robo areas, not engaging in inefficient competition in small-scale markets."

Global Timeline for Robotaxi and Robovan

Precisely based on such forecasts, Momenta's L4 scaled unmanned business is accelerating landing.

Regarding Robovan, the company has launched small-scale pilot operations in Xiangcheng, Suzhou. Partners cover China Post, JD.com, SF Express, ZTO Express, STO Express, YTO Express, J&T Express and other top express logistics enterprises. Application scenarios have expanded from courier connections to night delivery, fresh food, FMCG, etc. Management plans to complete commercial closed-loop verification in two core domestic cities by the end of 2026, expand to dozens of domestic cities in 2027 and enter the UAE market for the first time, and achieve nationwide scaled operations in 2028.

Regarding Robotaxi, Momenta has already conducted tests or operations in 6 cities in 3 countries. Domestically, licenses have been obtained in Shanghai, Suzhou, Wuxi, and Shenzhen; Overseas, besides Abu Dhabi road test licenses, in July 2026 it also became the first Chinese enterprise to obtain the Federal Motor Transport Authority of Germany's national L4 test license. The company plans to obtain test licenses in more than 10 cities globally by the end of 2026, with fleet size reaching hundreds of units; in 2027, expand to more than 30 cities, with fleet size reaching thousands of units.

Another path for going global is parallel to Chinese model exports and overseas localization projects. Currently, models equipped with Momenta's solution have entered 10 countries and regions such as Norway, UK, UAE, Singapore, Australia. With the gradual relaxation of European city NOA-related regulations in 2027, Momenta's accumulated overseas customer relationships and massive driving data are expected to accelerate its penetration in the international market.