Either define the rules, or step aside.

Author: Gu Qingyun

Source: Alter Talks Tech

To increase the "Mo content" in the automotive market, you must iterate on model parameters in the lab and build relationships with consumers under the spotlight.

At the 2026 Beijing Auto Show, the air still carried the scent of new car leather, but the underlying tone had changed.

Amid this noisy tech forum, Momenta CEO Cao Xudong made a stark, almost apocalyptic prediction: in autonomous driving, only 2-3 companies in China and 3-4 globally will rapidly emerge as winners, and the industry landscape will quickly consolidate.

As a provider with over 200 cumulative designated vehicle models, more than 70 delivered mass-production models, and over 800,000 units of intelligent driving assistance systems installed, Cao Xudong is confident that Momenta will capture one-third of the global autonomous driving market. At the Beijing Auto Show, he boldly stated: "Autonomous driving is the prelude to physical AI, and Momenta aims to be a platform player in this space."

Behind the grand narrative of "Physical AI," a troubling question has always hung over Momenta: at product launches by automakers such as GAC, SAIC, and Mercedes-Benz, Momenta is merely a small footnote on the PowerPoint slide—the “Momenta content” remains a topic automakers avoid.

01 The "Shadow Champion" of the First Tier: Strong in Algorithms, Weak in Mindset

Over the past two years, various autonomous driving reviewers and professional institutions have tested nearly all mainstream models on the market, producing a series of autonomous driving rankings.

A paradoxical phenomenon: On unofficial or semi-official autonomous driving competitions, vehicles equipped with Momenta’s solution frequently appear in the top tier, and in some complex urban NOA tests, its human-like dynamic performance has repeatedly outperformed Huawei’s ADS and Tesla’s FSD. Yet, in consumers’ minds, the top-ranked vehicles almost never carry the “autonomous driving” label.

A direct example is the second China Intelligent Driving Assistance Competition, Ningbo leg, organized by First Electric. The testing process was堪称 "hell mode": the entire route was concealed beforehand, featuring eight challenging checkpoints over a 29-kilometer course—including narrow residential roads, blind-spot U-turns, and artificial obstacles—along with deliberately difficult tasks such as identifying transparent plastic wrap.

When the final rankings appeared on screen, the entire automotive industry was stunned—because the top spot went not to問界, Li Auto, XPeng, or Avatr, but to Buick’s Zhijing L7.

Some have criticized it as an "unofficial ranking," while others have attempted to provide a technical explanation: The Buick Electra E5 is equipped with the Momenta R6 Flywheel large model, whose decision-making capabilities are trained on 4 billion kilometers of real-world data, offering significant advantages in complex scenarios. For instance, when navigating chaotic traffic during a left turn at a roundabout, its "perception-decision-execution" closed-loop response is 0.3 seconds faster than traditional systems...

This ranking ignited public debate but also exposed Momenta's issues.

First, there’s the technical jargon of "social isolation." Terms like reinforcement learning reward functions, feedback loops, and flywheel large models might elicit a nod from seasoned engineers but are just "academic gibberish" to average users. What we want is to "avoid erratic e-bikes like an experienced driver," not ask AI what an "end-to-end monolithic architecture" is.

Second, there is the issue of identity confusion among benchmark models. While the IQI L6 has delivered an impressive performance in intelligent driving, IQI’s marketing focus has consistently centered on the “Lizard Chassis” and the “Whole-Vehicle Intelligent Hub.” BYD and Yangwang have partnered with Momenta on a deeply co-developed solution, yet all the exposure and accolades have been attributed to “Sky Eye.”

Momenta is like a top chef cooking in prestigious restaurants, yet his name never appears on the menu. This disconnect between superior performance and weak recognition has left Momenta in the position of a "shadow champion."

According to Cao Xudong’s logic, the intelligent driving market is about to enter a淘汰赛 (winnowing phase), where only two or three suppliers will survive. The question is—how will automakers choose? Will they pick suppliers with superior technical specifications, or those that can drive sales? The answer is obvious.

At least for now, introducing Momenta’s leading capabilities to consumers often requires lengthy explanations starting with the founding team’s background and citing one technical term after another. In contrast, Huawei’s Qiankun needs only one sentence—“Our intelligent driving is from Huawei”—to create a psychological impression of “greater safety” in users.

02 The "soul theory" has not disappeared; the "Mo content" deliberately diluted by automakers

Momenta has struggled to establish a "brand benchmark" because, fundamentally, it chose the hardest and most solitary path: building the underlying operating system for global OEMs.

In the automotive industry, where "brand sovereignty" is paramount, carmakers hold complex feelings toward powerful suppliers—akin to a secret affair blending dependence, suspicion, and anxiety. They want suppliers’ technology to fill gaps in intelligence and connectivity, yet fear being overshadowed and, above all, losing their core identity.

The veil named "co-research" is the first barrier preventing "molybdenum content" from becoming visible.

Taking SAIC as an example, as Momenta’s most critical foundational partner, multiple models including IM Motors and Volkswagen have integrated Momenta’s autonomous driving solutions, yet consistently downplay the supplier’s role in their marketing—for instance, IM Motors has branded its autonomous driving system as “IM AD.”

In the automotive industry, such practices have almost become an unwritten rule: money can be given, data can be exchanged, but brand premium cannot be shared. Automakers need to demonstrate to capital markets and consumers that they possess "full-stack in-house R&D" capabilities, leading to the deliberate dilution of "Mo's content."

Before multinational giants like BBA and Toyota, this dilution is even more apparent.

While Qiankun Intelligence is accelerating its penetration into the traditional fuel vehicle market through collaborative models such as the Audi Q5L, Momenta has been knocking on the doors of Mercedes-Benz and BMW’s R&D centers for years.

The established giants chose Momenta for its global mass-production capabilities and low-cost adaptation solutions; in their view, intelligent driving is merely a feature on a spec sheet. You’ll see “Superior Safety Assistance” on Mercedes’ promotional materials, but never “Powered by Momenta.”

In contrast, Huawei's Qiankun has broken the curse not only through its technology but also by providing automotive manufacturers with the traffic and brand credibility needed to revive their businesses. Partners like Seres and BAIC are willing to “surrender their soul” in exchange for survival, because consumers buy AITO and SHENGYE not just for the cars, but for the “Huawei quotient” they represent.

By 2026, Huawei's Qiankun had accelerated its market expansion: At the launch event themed “Integrating Intelligence into Every Vehicle,” Huawei Qiankun adopted an almost saturation-style approach, collaborating with GAC to launch Qijing, partnering with Dongfeng to introduce Yijing, and working with SAIC General Motors Wuling to unveil Huajing.

In particular, Hua Jing largely means that Huawei's Qiankun driver assistance and intelligent cockpit systems will no longer be limited to the mid-to-high-end market above 200,000 yuan, and are likely to extend down to the 150,000-yuan market segment by 2026—the current price range targeted by Horizon Robotics and Momenta.

As a purely technical platform, Momenta currently lacks strong bargaining power for driving sales.

If automakers feel that using your product is merely for convenience and ease of use, and that associating their brand with yours could dilute their own brand value, the only outcome of this negotiation is that you become the "thermal underwear" of autonomous driving—close-fitting, life-saving, and indispensable, yet never in the spotlight.

On the positive side, even in a supporting role like "thermal underwear," Momenta's data volume continues to grow, and its technological foundation becomes increasingly solid. The risk lies in the fact that if "Moore's content" cannot become a tangible, user-driven value proposition akin to "Intel Inside," it remains vulnerable to being replaced at any time by automakers.

Chapter 03: The One-Sided Storyteller: When Elitism Meets Traffic Logic

In the second half of the smart driving industry, success hinges on the depth of algorithms, but victory often goes to those with the broadest brand persona.

Hongmeng Smart Drive has Yu Chengdong, who excels at turning dry technology into a super IP of “far ahead” and “changing the world”; Xpeng Auto has He Xiaopeng, who has established a persistent image of a tech enthusiast who never compromises; Li Auto has Li Xiang, who precisely targets the safety concerns of family users with the empathy of a product manager.

The founding team of Momenta is a typical group of technical elites.

Cao Xudong, frequently active in top academic forums and research centers, is known for his minimalist, pragmatic, and calm style, which may be perceived as "boring" in the consumer-driven automotive market.

Reviewing Cao Xudong’s public speeches over the past two years, he continues to try to win over audiences through “logical deduction”: discussing R7’s reinforcement learning world model, algorithmic convergence rates, and decreasing marginal costs. While these topics are academically impeccable, they represent a “disaster” in mass communication.

However, Cao Xudong is clearly trying to break beyond the jargon-heavy context and connect with an increasingly larger audience.

During the Beijing Auto Show, it not only predicted that "the industry landscape will rapidly converge," but also proclaimed, "We hope to write the legend of an Eastern Silicon Valley alongside all Chinese AI companies."

The sense of dissonance lies in applying the rigorous logic of linear algebra to promote a viral restaurant. The so-called "shocking claims" often ring hollow, lacking concrete, relatable emotional connections, resulting in much ado about nothing.

To some extent, Cao Xudong might benefit from learning from Yin Qi of Qianli Technology: focus on the technology he excels in and leave the spotlight to more articulate figures like Zhao Ming—within a single product launch, Zhao Ming generated significant visibility for Qianli Intelligence Driving by asserting claims such as “putting Tesla in its place” and “becoming the leader in intelligent driving within three years.”

Keep in mind that Qianli Technology's intelligent driving revenue in 2025 was only RMB 350 million, accounting for less than 4% of total revenue. Even with Geely Group as a major shareholder, an intelligent driving company that has just completed initial validation is unlikely to make a significant impact in the industry.

Qianli Zhijia's breakout and its ambition to equip 8 million vehicles within three years precisely confirm the shift in the underlying logic of intelligent driving: it is no longer just about being an automaker's Tier 1 supplier, and there is no longer a comfort zone from the Bosch era. Those who move too slowly aren't just left behind—they are eliminated outright.

The pride of technical elites often lies in believing that "a great product speaks for itself," but in the 2026 autonomous driving battleground, a great product that can't speak for itself will likely be drowned out by the noise. In today’s highly fragmented information environment, failing to capture user attention within three seconds is a failed marketing effort.

For Momenta, increasing the "Mo content" in the automotive market requires not only iterating model parameters in the lab but also building relationships with consumers under the spotlight. After all, in the final stage of market consolidation, only brands that are remembered have the right to survive and negotiate.

04 Final Thoughts

In a consolidating market, there is no middle ground—either you set the rules, or you step aside.

For Momenta, the deployment of 800,000 vehicles is both a source of confidence and a warning: it cannot remain merely a supporting player in sales narratives—it must strive to become the primary choice that drivers are willing to pay for. As Huawei’s Kunlun system has already entered the premium fuel vehicle market and continues to expand into lower-tier markets, the window of opportunity for Momenta to prove its “motor intelligence” is closing.

Between the vanished heroes and the visible backbone lies not just algorithms, but also user reputation and perception.