Gasgoo Munich- In June, Seres officially launched its new automotive brand, AIVA. Its predecessor was Seres' Landian; behind it stand Seres, ByteDance's Volcano Engine, CATL, and Chongqing state-owned assets. At the launch, AIVA declared its slogan: "AI defines the car. AI comes first, then the car."
AIVA’s launch comes at arguably the best moment in Seres' history, with AITO dominating the market and Huawei’s empowerment clearly visible. Yet at this juncture, Seres chose to "start fresh" — rolling out a brand independent of the AITO and Huawei ecosystems.
Carrying 6.671 billion yuan in capital restructuring and high expectations, what real possibilities can AIVA bring to Seres?
How Substantive Is AIVA’s New AI Narrative?
For a new brand to stand firm, it must first answer one question: Why should it be remembered? AIVA’s answer: "AI defines the car."
Image source: Saidou Technology
AIVA President Li Bo summed up the brand’s core philosophy at the launch: "AI comes first, then the car." Distinct from the industry mainstream of "building the car first, then adding intelligent features," AIVA advocates establishing AI as the vehicle’s core "brain" from the R&D source. Volcano Engine VP Yang Liwei elaborated on stage: "Our understanding of an AI car isn't just putting AI into a vehicle, but making the car a new species of physical AI. If a car is defined around AI from day one, its interaction methods, intelligence ceiling, and user experience will fundamentally change."
Thus, in AIVA’s vision, AI will no longer be a "bolted-on" function module, but a native capability of the vehicle — capable of continuous iteration and autonomous evolution. The first mass-production model, the AIVA ME7, will launch in 2026, with the entire lineup targeting the mainstream market above 200,000 yuan.
AIVA isn’t the first automaker to talk about AI. Tesla talks about it, as do XPENG and Li Auto. But AIVA’s approach differs from them all.
Image source: @GraceTaoLin-Tesla
In early 2026, Tesla VP of External Affairs Grace Tao stated clearly in an interview that Tesla is no longer just an EV company, but has transformed into a tech enterprise centered on AI, robotics, and energy.
On the Q1 2026 earnings call, Tesla further clarified its strategic positioning: "FSD is the core product; the car is merely the delivery vehicle." Behind this statement lies a clear logic: AI exists before any single hardware. The car is just the first stop for AI capability expansion, while the humanoid robot Optimus represents the next application scenario.
After establishing this new positioning, Tesla adjusted its automotive business. That same year, Tesla announced it would gradually halt production of the Model S and X, shifting corresponding factory capacity to mass production of humanoid robots, and planned to raise 2026 capital expenditures sharply to $25 billion, going all-in on AI and robotics.
Looking at domestic new forces, XPENG upgraded its positioning in 2025 to a "mobility explorer in the physical AI world," planning to lift physical AI R&D investment to 7 billion yuan in 2026. Li Auto, meanwhile, explicitly proposed shifting from "creating a mobile home" to the field of embodied intelligence in early 2026, fully laying out humanoid robots and system-level general agents.
XPENG and Li Auto’s AI narrative logic aligns with Tesla’s — the car is the carrier for AI. The perception, decision-making, and control capabilities accumulated on top of it migrate to broader scenarios like robots, with the primary focus often landing on smart driving.
AIVA, however, focuses on the Doubao large model empowering the smart cockpit, while adopting DeepRoute’s smart driving solution.
Image source: Ji Yue Auto
When discussing brands with an AI narrative, one must mention Ji Yue Auto, jointly built by Baidu and Geely. It also entered the market with an AI positioning, proposing the concept of a "car robot" and attempting to capture the high-end market through smart driving technology. However, the result is obvious: the advanced concepts Jiyue preached were not recognized by consumers, and the market response was the most blunt evidence.
Monthly sales of the Jiyue 01 and 07 lingered in the hundreds for a long time. In the first half of 2024, Jiyue’s cumulative sales were only 2,700 units. By the end of 2024, full-year deliveries fell short of 14,000 units. In December 2024, Jiyue faced a cash crunch, and the company ceased operations immediately.
For AIVA, the 200,000-plus yuan market it is targeting is already one of the most fiercely competitive. Even with lessons from the past, a new brand breaking through in this price segment still faces the challenge of building brand awareness from scratch. In this regard, an analyst at Gasgoo Auto Research noted: AIVA lacks Huawei’s endorsement and Xiaomi’s mass base, so building trust takes time. Additionally, whether AIVA’s AI narrative can translate into a tangible, easy-to-use differentiated experience for users in mass-produced cars remains an unknown for now.
In short, for all the AI narratives, we must eventually return to a simple test of validity: every definition written on a PPT will be re-examined upon delivery; the distance between concept and fact can only be measured on the mass-produced car. AIVA is no exception.
Can AIVA Recreate an "AITO-Style Myth"?
AIVA has told a narrative logic about AI. Whether this narrative can bridge the gap from the launch stage to the mass-produced car, and truly transform into a product experience users are willing to pay for, is perhaps the true yardstick for measuring AIVA’s value to Seres.
Like AITO, AIVA is a deep collaboration between an OEM and a top-tier domestic tech company. But it differs significantly from AITO in terms of cooperation depth, Seres’ control rights, and the market entry window.
Before the AIVA brand was officially unveiled, rumors of "ByteDance building cars" had been fermenting in the market for nearly a year. At the end of July 2025, talk of "ByteDance preparing a Doubao Car" first circulated, to which Volcano Engine quickly clarified, calling the online "Doubao Car" rumors purely baseless.
Even so, when the brand name "Saidou" appeared in public view, the association between "Sai" and "Doubao" was almost inevitable. Many voices interpreted this as the product of a deep fusion between Seres and Doubao, even calling it directly "Doubao Car."
Image source: ByteDance
On June 6, 2026, ByteDance released a statement: it has no plans to build cars or launch a car brand, and "Saidou" is not a car brand launched by ByteDance or Doubao, with no equity cooperation between the two. Volcano Engine’s role was clearly defined as a technology supplier — providing the Doubao large model, smart cockpit, and other technical services to help partners improve in-car intelligent interaction experiences.
At the same time, reports citing insiders revealed that Volcano Engine’s involvement in AIVA this time exceeds the level of cooperation seen previously with SAIC Roewe’s "joint definition, joint development" — going even deeper. Although ByteDance is not building cars, it is participating as a deep technical partner.
Regardless of how ByteDance’s role is defined, one fact is certain: AIVA is a new brand Seres is building by introducing AI capabilities from another tech company, separate from Huawei. So, given the current variables and market backdrop, what can AIVA actually bring to Seres? Is there a chance to recreate another "AITO-style myth"?
Reviewing the rise of AITO, its success factors are very clear: Huawei provided full-stack deep empowerment from chips to operating systems, and from smart driving solutions to brand marketing; Seres Holdings took the lead, with both parties’ interests highly aligned.
Between 2021 and 2022, the Huawei formula was still scarce. This formula pushed AITO from zero to 420,000 annual sales within three years, directly driving Seres to turn losses into profits.
Image source: AITO Auto
However, compared with AITO, based on currently disclosed information, there are significant differences between the two brands.
First is the depth of participation. Huawei provided AITO with a deep bond of "full-stack delivery plus brand endorsement"; ByteDance has drawn a clear line with Saidou through official statements, clarifying Volcano Engine’s role as providing technical services like the Doubao large model and smart cockpit. Meanwhile, Volcano Engine’s Doubao model already covers over 50 automotive brands and 145 models on the market, meaning if AIVA wants to differentiate, it must go further in integration depth than other Doubao partners.
Second is Seres’ control. In the AITO system, Seres Holdings holds and leads operations; in AIVA’s equity structure, Seres has retreated to the second-largest shareholder, with Chongqing state assets controlling the stake, and Seres is merely a participating party. Third is the market window. When AITO was born, the high-end new energy market was still a blue ocean, while the 200,000-plus yuan market AIVA is entering is already a fiercely competitive red ocean.
Furthermore, the subsequent development of the AI narrative logic AIVA is pursuing is currently also a question mark.
Even so, AIVA’s strategic value to Seres remains clear.
First, for Seres, AIVA’s most intuitive value lies in forming a staggered complement with the high-end-focused AITO, helping Seres cut into the larger mainstream mass market.
Second, looking at the equity structure, Seres has retreated to the second-largest shareholder, and AIVA’s operations are independent of the AITO system, with its operating performance not included in the listed company’s consolidated statements. This structure allows Seres to avoid the risk of relying entirely on a single partner in its brand layout, while also limiting the impact of AIVA’s early-stage investment on Seres’ overall financial performance.
Third, in terms of technology roadmap, it accumulates different intelligent capabilities for Seres — AITO takes the HarmonyOS ecosystem route, while AIVA takes the Doubao large model route. Pursuing a dual-track strategy gives Seres a layout in multiple technology paths.
Additionally, through the AIVA platform, Seres has formed a deep industrial alliance with CATL and Chongqing state assets. If this mixed-ownership model works, it can be replicated to more business areas in the future.
Seres’ layout for AIVA follows simple logic: success means complementing AITO and adding another leg; failure means controllable losses.
Seres Needs AIVA
Seres got on its feet thanks to AITO. But as AITO becomes more successful, one point needs to become clearer: Huawei is becoming Huawei for more people, and Seres needs to find its own voice.
In 2025, AITO’s annual sales were about 426,000 units, occupying the absolute core position in Seres’ overall sales. But behind the excellent sales data lies a fact that cannot be ignored.
According to data disclosed by Seres, in the first half of 2025, Seres paid 20 billion yuan in procurement fees to Huawei, accounting for about one-third of total revenue over the same period. Looking back over three and a half years, Seres has paid over 75 billion yuan in procurement fees to Huawei. Based on the approximately 147,000 AITO deliveries in the first half of 2025, about 136,000 yuan flows to the Huawei system for every AITO sold.
The industry jokingly refers to the procurement fees Seres pays to Huawei as the "Huawei Tax."
The high "Huawei Tax" directly erodes Seres’ profit margins. In 2025, Seres’ annual operating revenue reached 165.054 billion yuan, up 13.69% year-on-year, a record high; but net profit attributable to shareholders of the listed company was only 5.957 billion yuan, a mere 0.18% year-on-year increase. With revenue soaring but profit growth negligible, the dilemma of "increasing revenue without increasing profit" limits Seres’ profit space.
Image source: HarmonyOS Mobility
Meanwhile, HarmonyOS Mobility is empowering more automakers, and the "Five Realms" pattern has taken shape. The scarcity of AITO as HarmonyOS Mobility’s first brand is being diluted. Seres urgently needs a front independent of Huawei to hedge against the long-term risks of "Huawei dependence syndrome."
In the view of analysts at Gasgoo Auto Research, Seres’ dual-line layout is clearly reasonable: AITO attacks the high-end market; AIVA attacks the 200,000-yuan class market. Seres focuses on AITO; Saidou Technology independently operates AIVA, each performing its own duties. Over-binding to a single ecosystem is itself a risk, and Seres’ dual-line parallelism helps it jump out of the homogenized competition within the Huawei system.
They also further stated that if AIVA doesn’t work out, it could have a certain impact on Seres’ brand image and capital market confidence — the market might question Seres’ ability to "start fresh" outside of Huawei.
It is well known that this is not Seres’ first attempt at a dual-brand strategy. As early as 2023, Seres launched the Landian brand, positioning it in the 100,000 to 150,000 yuan mass market, intending to complement the high-end AITO.
But this strategy ultimately failed to materialize. Landian’s sales were only about 34,000 units for all of 2024, slipping further to about 20,000 units in 2025, and by early 2026, monthly sales were less than 500 units.
Industry analysts pointed out that its sales were far from the break-even line, long dragging on group resources, with the product cycle entering its final stage and sales turning downward — this became the direct quantitative signal triggering this asset restructuring.
Rather than patching up a continuously loss-making brand internally, it is better to conduct a thorough restructuring externally. This is the direct driving force behind Seres pushing AIVA.
However, compared with Landian, Seres no longer holds a controlling stake in AIVA. If AIVA succeeds in the future, Seres can only share part of the proceeds as a financial beneficiary, struggling to become the brand leader as it did with AITO.
Conclusion
Seres made the move of AIVA during its peak period. AITO proved it has the ability to succeed with Huawei, while AIVA has to answer another question: leaving Huawei, can it still walk a path alone?
Whether this path can be walked through depends not on the definitions at the launch, nor on AIVA’s AI narrative, but on the yardstick: after the AIVA ME7 is delivered, can users perceive this set of AI-defined product capabilities in real driving scenarios?
