AI・テクノロジー
In-Vehicle AI Shock | Alibaba Qwen Becomes Standard in 9 Chinese EV Brands
機械翻訳 / Machine-translated
A
AI Friends
2026年4月29日・49分で読める
機械翻訳 / Machine-translated
@aifriends
AI Friends（https://aifriends.jp）のクロスポスト公式アカウント。AIツールの紹介・使い方・できることを、中学生でもわかるやさしい日本語で届けます。
"While gripping the steering wheel, order dinner delivery and book a hotel at your destination—just by speaking." — What sounds like a sci-fi movie is already becoming reality in Chinese EVs.
On April 24, 2026, the opening day of the 2026 Beijing Motor Show, Alibaba made a bombshell announcement.
"How will Toyota and Honda even compete?" "Can you do the same things in a Japanese car?" "Is privacy okay?" "How is the battle for dominance among China's AI tech giants playing out?" — We break down all the answers you're curious about, in plain language anyone can understand.
First, let's organize what happened from three angles.
On April 24, 2026, the opening day of the 2026 Beijing Motor Show—the world's largest auto show, boasting 380,000 square meters of exhibition space—Alibaba made a headline-grabbing announcement.
Think of it like "a powerhouse school suddenly unveiling a new uniform at the opening ceremony of a sports festival"—that's the kind of attention it drew.
The announcement: Alibaba's proprietary AI model "Qwen" will be standard-equipped in vehicles from nine major Chinese automakers.
"Qwen" is a large language model (LLM—an AI that can understand and generate text like a human) developed by Alibaba. As of 2026, it has surpassed 1 billion downloads worldwide and is released under the Apache License 2.0, which allows free commercial use.
Think of it as "putting a super-capable, freely available AI directly into a car as standard equipment."
Alibaba Cloud (Alibaba's cloud business) and automakers are integrating Qwen directly into dashboards.
While voice assistants have been installed in individual cars before, having a ChatGPT-class LLM as standard in-vehicle equipment is a major turning point for the industry.
International media have described it as "the evolution from car navigation to AI butler."
The nine manufacturers that announced Qwen integration are as follows:
① BYD (world's largest EV maker), ② Geely (parent company of Volvo), ③ Li Auto, ④ Changan, ⑤ Dongfeng, ⑥ BAIC, ⑦ Great Wall Motor, ⑧ SAIC Volkswagen (joint venture between SAIC and Germany's Volkswagen), ⑨ SAIC IM Motors.
It's the equivalent of "most of the major league of Chinese EVs all switching to the same AI at once."
China's EV market in 2026 exceeds 13 million units annually, accounting for roughly 60% of global EV sales—a massive market.
The combined sales of these nine companies cover more than half of the Chinese EV market, and the impact is comparable to "Toyota, Honda, Nissan, Suzuki, Mazda, and other major Japanese automakers all adopting the same AI simultaneously."
Particularly notable is the inclusion of SAIC Volkswagen, a joint venture with a foreign automaker.
This makes it clear that "even foreign manufacturers in the Chinese market have no choice but to rely on Chinese-made AI."
In 2025, BMW and Alibaba already announced a deepened partnership on in-vehicle AI, and Qwen is becoming an indispensable part of foreign automakers' China market strategies.
Before the simultaneous announcement by nine companies, FAW Group's Hongqi brand had already installed Qwen in its PHEV model, the HS6.
It's like "an ace player debuting ahead of the match and already delivering results."
The positive reception of Qwen's implementation in the Hongqi HS6 helped pave the way for the large-scale rollout to nine companies.
Think of it as "the drug's efficacy was confirmed with the first patient, so full-scale prescriptions are now beginning."
Hongqi is FAW Group's luxury brand and is also used as an official vehicle by China's top central leadership.
Its positioning is like "a brand that's both Japan's Lexus and the go-to car for government agencies."
With a proven track record on such a flagship brand, other manufacturers can adopt it with confidence.
Hongqi HS6 drivers simply say things like "Book one night at the Hilton in Shanghai, starting tomorrow at 7 PM" or "Order two portions of Peking duck that can be delivered within 30 minutes," and the in-vehicle AI completes the arrangements automatically.
The experience of "no longer needing to touch your phone while driving" has earned enthusiastic support from Chinese users.
Qwen's in-vehicle AI offers four representative functions:
① Food delivery orders (integrated with apps similar to Chinese Uber Eats), ② Hotel reservations (integrated with travel sites like Trip.com), ③ Purchasing tickets for tourist spots (theme parks, museum entry tickets), ④ Package tracking (checking delivery status via China Post, SF Express, etc.)
Imagine a world where "if you feel a bit hungry while driving, you just ask the AI and delivery arrives before you get home."
If traditional car navigation is "a companion who tells you where to go," Qwen's in-vehicle AI is "a butler who handles your entire life on the move."
Think of it as "car navigation + secretary + concierge all in one service."
Where before you could only give voice commands like "turn right at the next signal," the AI now autonomously handles complex, multi-step tasks.
The hallmark feature is being able to say things like "Book the cheapest travel and hotels to and from Shanghai for a business trip in three days," and having it all taken care of.
The accuracy of voice command comprehension depends on Qwen's natural language processing capabilities. As of 2026, Qwen supports over 100 languages including Chinese and English and has reached a level capable of handling expert-level conversations.
The technical makeup of Qwen's in-vehicle AI is top-tier in the industry:
① Runs on NVIDIA automotive chips (DRIVE series), ② Hybrid design combining on-device processing (handled entirely by the car's chip) and cloud processing (via Alibaba Cloud).
It's as smart as "automatically switching between home Wi-Fi and your phone's 4G to always maintain the best connection."
Basic functions work even in environments with unstable network connections, such as underground parking lots or tunnels.
It's the reassuring design of "being able to enter expense records offline in a household budget app even when the signal is poor."
On the cloud side, Alibaba Cloud's massive computing infrastructure executes multi-step task plans in real time.
When asked to "book the Hilton in Shanghai," the system ① checks room availability at Hilton Shanghai → ② compares rates for the desired dates → ③ accesses the booking site via API → ④ completes the reservation → ⑤ sends a confirmation email, all in under a second.
It's the efficiency of "a task that would take a new secretary 30 minutes, completed instantly by an AI butler."
The NVIDIA DRIVE series is the global standard chip for in-vehicle AI as of 2026, adopted by major manufacturers including Tesla, BMW, and Mercedes, and its combination with Qwen delivers both performance and versatility.
With an AI that delves this deeply into daily life, concerns naturally arise: "Won't all the driver's preferences, behavioral history, and payment information be completely exposed to Alibaba?"
Alibaba Cloud explains: "Personal information is processed with encryption, and anonymization is also applied on the vehicle side."
Think of it as "conveying an order to the kitchen without revealing the customer's name before bringing out the food."
However, under China's National Intelligence Law, tech companies are obligated to submit data to the government upon request, meaning "the same level of personal protection as Europe's GDPR or Japan's Act on the Protection of Personal Information cannot be expected"—that's the reality.
Wariness toward Chinese-made AI is growing in the US and Europe, with regulations on Chinese apps like TikTok and WeChat expanding.
There's also "the risk that Qwen integration could actually become a reason EVs don't sell when Chinese EVs are marketed in the West."
How Qwen will be handled in Geely-owned Volvo vehicles and SAIC Volkswagen's export models for the European market is a major challenge going forward.
The likely outcome is "installed in the Chinese market, switched to a different AI for Western markets"—a two-tiered approach.
In Japan too, there is an emerging possibility that import restrictions on Chinese AI-equipped vehicles could be debated due to privacy concerns.
Tencent is countering with in-vehicle AI that integrates SoundHound Chat AI Automotive.
SoundHound is a US-based voice AI company, with strong support in English-speaking markets.
The contrast is like "Alibaba going the self-sufficient route with Qwen, while Tencent brings in outside specialists."
Tencent Intelligent Mobility integrates in-vehicle entertainment, maps, and communication functions.
The emerging dynamic looks like "Alibaba dominating domestically in China, Tencent having the advantage for overseas exports."
Tencent's strength lies in WeChat integration—a structure where China's national app used by over 1.2 billion people domestically works in tandem with in-vehicle AI. It's easy to understand if you "imagine a Japanese car with complete LINE integration."
As of 2026, Tencent's in-vehicle AI is installed in approximately 500 vehicle models, deployed in over 10 million new cars annually.
Industry analysts describe Alibaba's simultaneous nine-company announcement as "a counteroffensive to chip away at Tencent's first-mover advantage."
Baidu released ERNIE 5.0 on January 22, 2026—a native omni-modal model (integrating text, image, voice, and video processing) with a 2.4 trillion parameter MoE (Mixture of Experts) architecture.
It boasts the strength of being "a super-large AI several times the scale of Qwen."
It has achieved the distinction of ranking #1 in China and #8 globally on the text benchmark "LMArena."
Baidu's strength lies in integration with its autonomous driving platform "Apollo," offering a genuine autonomous driving solution where "the AI handles not just the assistant functions, but the driving itself."
It is also operating robotaxi (autonomous taxi) services in major Chinese cities including Beijing and Wuhan.
The dynamic is one where "Alibaba focuses on in-car services, Baidu focuses on the driving itself"—they're attacking at different layers.
Baidu has established a unique position by "achieving Tesla-level autonomous driving technology through Chinese domestic development."
As of April 2026, Baidu Apollo has conducted test drives in over 150 Chinese cities, with commercial robotaxi operations running in more than 20 cities.
In April 2026, Huawei announced the start of mass production of the Ascend 950PR chip, with a production plan of 750,000 units annually.
This is a strategy of "using a proprietary AI chip designed to rival NVIDIA's H100 while circumventing US export controls."
Think of it as "in-house manufacturing of the parts needed to repair a home appliance"—a path of independence.
Huawei also operates its own EV under the AITO (Wenjie) brand, achieving full vertical integration from in-vehicle AI to driving control.
The four companies each have their own color: "Alibaba for AI, Tencent for communications, Baidu for driving, Huawei for everything."
The strength of being able to build a proprietary AI semiconductor ecosystem even under US sanctions is a major long-term differentiator.
As of 2026, Huawei's Ascend chips have secured a major order worth $5.6 billion (approximately 840 billion yen) from ByteDance, and its presence in AI semiconductors continues to grow.
With the comprehensive capabilities of "China's combined Intel and NVIDIA," Huawei is carving out its own distinctive position in the in-vehicle AI market as well.
Toyota filed 473 AI-related patents over the five years from 2019 to 2023, five times the 92 filed in the preceding five years.
The determination is like "increasing R&D speed fivefold to compete with China."
In a ranking of AI readiness among 20 global automakers, Toyota ranks 3rd in the world by number of patents held.
Its strength: "Not losing to Chinese EV players when it comes to patent count."
The strategic pillar is multi-agent development—a concept in which multiple specialized AI agents collaborate to support the driver.
Think of it as "not one all-purpose secretary, but a three-person team: one for driving, one for navigation, one for reservations."
As of 2026, Toyota's cumulative five-year investment in autonomous driving-related areas stands at approximately 1.5 trillion yen.
The current state: "The money and ideas are in place—it all comes down to implementation speed."
A key challenge is falling behind on the shift to SDV (Software Defined Vehicle—a car whose value is determined by its software).
"Can a Japanese manufacturer good at hardware win on software?" is Toyota's defining challenge.
In April 2026, Honda announced a one-year delay in the market launch of AI autonomous driving, pushing it from the original 2027 to 2028.
The ripple effects of a revised EV strategy have pushed AI development back as well.
It's a cautious posture like "a player injured just before the game taking a longer recovery period."
Partnership strategy: ① Collaborating with China's Momenta on ADAS (Advanced Driver Assistance Systems), ② Collaborating with US-based IBM on SDV.
It's a pragmatic judgment that says "domestic Japanese development alone can't keep up, so we're hiring specialists from China and the US."
Momenta is a Chinese autonomous driving startup known for technology on par with Tesla.
The structure of "putting a Chinese brain in a Japanese car" raises the possibility of future collaboration with Qwen in-vehicle AI.
As of 2026, Honda plans to convert over 50% of its Chinese market sales to EVs and PHEVs.
The future of "having no choice but to adopt Chinese-made AI in order to survive in the Chinese market" is becoming increasingly real.
Industry analysts say the keys to Nissan's comeback from its ongoing restructuring are "AI" and "the Chinese market."
It's a situation like "a baseball team on a losing streak rebuilding with a new manager and foreign players."
Management talks with Honda are also ongoing, and "a future where a Nissan-Honda alliance adopts Chinese-made AI" is one possible scenario.
As of 2026, Nissan is struggling in the Chinese market, with its share falling below 10%.
"Reclaiming ground in the Chinese market requires building cars tailored to Chinese users' preferences."
If Qwen in-vehicle AI continues to standardize in the Chinese market, Nissan will have little choice but to follow suit.
Related reports indicate that in the second half of 2026, Nissan is expected to announce Chinese-made AI integration in a new EV model for China.
The day when "a Japanese-branded car with a Chinese brain" becomes the norm is drawing near.
The concept of a Toyota-Honda-Nissan alliance in Japan is also drawing attention as a potential catalyst to accelerate the AI shift.
Sato-san frequently travels to Shanghai as a China representative for a major trading company.
In June 2026, the moment he gets into a BYD rental car, Qwen in-vehicle AI greets him: "Welcome back, Sato-san. For tomorrow's meeting in the Pudong area, your reservation at the Hilton Shanghai is confirmed."
It's a futuristic feeling where "your hotel for the business trip is already sorted the moment you get in the car."
When Sato-san says "I'd like to eat hairy crab for dinner," Qwen instantly suggests three well-reviewed specialty restaurants, and he completes a voice reservation for his preferred one.
Sato-san is stunned by the experience of "being able to handle transportation, meals, and accommodation all on your own while driving, without ever touching your phone."
In 2027, Sato-san's company formalizes an internal rule: "Rental cars for China business trips should prioritize Qwen-equipped vehicles."
Time efficiency before and after business negotiations improves dramatically, enabling him to increase business trips from two to three per month.
It's an example of "an AI assistant-equipped car amplifying a businessman's capabilities."
Sato-san is promoted to Shanghai branch manager in 2027, with his annual salary rising from 12 million yen to 18 million yen.
He is spreading the lesson within his company: "Mastering Chinese-made AI early is an essential skill for success in Chinese business."
Yamada-san works in Toyota Motor Corporation's DX Promotion Division, developing next-generation in-vehicle AI.
In May 2026, his division manager returns from the Beijing Motor Show with an urgent report: "I experienced the Qwen in-vehicle AI in an actual car. This is seriously incredible."
Yamada-san immediately assembles a counter-strategy team and, within three weeks, drafts a prototype plan for "Toyota's original AI assistant."
It's the agility of "a castle lord who immediately forms a defense unit the moment the enemy attacks."
Strategy: ① Consider using Chinese AI in tandem for the Chinese market, ② For the domestic market, promote Toyota's proprietary AI on quality and safety, ③ Standardize Tesla-level SDV features for the US/European markets.
Flexible enough to "adapt the optimal AI strategy for each market."
In 2027, Toyota equips the new Crown with a domestically produced AI called "TOY-AI (working title)" and offers it as a paid service at 3,000 yen per month.
First-year subscribers: 100,000 people, generating 3.6 billion yen in new annual revenue.
Yamada-san leads "the transition from one-time hardware sales to a software-based recurring subscription model."
In 2028, Yamada-san is promoted to General Manager of the DX Promotion Division, with his annual salary rising from 11 million yen to 17 million yen.
It's a transformation model where "the threat from Chinese EVs became the catalyst to accelerate DX at a Japanese manufacturer."
Suzuki-san works at an IT company in Tokyo.
In August 2026, drawn by the weak yen and the high quality of Chinese EVs, he purchases a BYD ATTO 3 (a parallel import of a model not officially sold in Japan).
After delivery, he asks Qwen's in-vehicle AI "Do you support Japanese?" and is met with a fluent Japanese response: "Of course, Suzuki-san."
He is surprised by the high level of quality—"a Chinese-made AI that can hold a normal conversation in Japanese."
A month later, Suzuki-san starts relying on Qwen's in-vehicle AI for everything: commuting, shopping, and get-togethers with friends.
He delegates his entire life to the AI butler with requests like "Book a convenience store bento pickup for tomorrow morning" and "Plan a trip to Izu next weekend with hotels and a sightseeing route at the lowest prices."
It's a new experience that feels like "Siri or Alexa is not at home, but in the car."
Suzuki-san starts posting an "Experience Report on Chinese EVs + AI Butler" on his personal blog, which grows into a popular site with 500,000 monthly page views.
In 2027, his side income exceeds 300,000 yen per month, bringing his total annual income to 10.6 million yen combined with his primary salary of 7 million yen.
"People who can disseminate the Chinese EV user experience in Japanese are rare," so Suzuki-san finds success in this new niche.
If China's EV market in Japan truly expands after 2028, demand for "Chinese EV evangelists" like Suzuki-san is expected to surge.
A. As of April 2026, there are no confirmed plans for Qwen integration in Chinese EVs officially sold in Japan.
BYD is selling multiple models in Japan in 2026, such as the ATTO 3, but the Japan-spec versions feature a proprietary AI assistant.
The current situation is "different versions are released for the Chinese and Japanese markets."
With parallel imports, it may be possible to drive a Chinese-spec vehicle as-is, so some Chinese EV enthusiasts are experiencing it through personal imports.
From 2027 onward, if Chinese EVs gain market share in Japan as well, official sales of Qwen-equipped models may be considered.
"If you want to experience it right now, the quickest way is to rent a car during a business trip to Shanghai or Beijing."
BYD and Geely vehicles can be reserved through major rental car companies such as SAIC Rental and Shenzhou Rent-a-Car.
"Rental fees start from approximately 8,000 CNY (about 160,000 yen) for three days," and more people are trying it out on business trips.
A. Honestly, Chinese-made AI does carry personal information risks.
Under China's National Intelligence Law, companies are obligated to submit data upon government request, so the reality is that "the same level of strict protection as Europe's GDPR or Japan's Act on the Protection of Personal Information cannot be expected."
Business people and civil servants who handle confidential information should be cautious about using it for work-related calls and reservations.
The golden rule is "choosing what you ask the AI when making requests before or after an important meeting."
For personal use (meals, tourism, shopping), the risk is limited.
Think of it as "roughly the same level of data collection as a Google Home or Amazon Alexa at home."
If you place a high value on confidentiality, it would be wise to refrain from business use and limit it to personal purposes only.
In the second half of 2026, the Japanese government is reportedly moving to establish "guidelines for civil servants using vehicles equipped with Chinese-made AI," and it is highly likely that usage restrictions will expand in professional settings.
A. For some models sold in the Chinese market, the likelihood is growing.
The pragmatic judgment is that "selling in the Chinese market means equipping cars with AI that Chinese users are accustomed to will sell better."
Honda has already partnered with China's Momenta on advanced driver assistance systems, and "cars that are Japanese-branded but have a Chinese brain" are already in the making.
Toyota is also exploring collaboration with local AI companies for its Chinese market offerings.
However, for models sold in Japan and in Western markets, Toyota and Honda are expected to adopt their own proprietary AI, or AI from Google, Apple, or Microsoft.
"Adapting the optimal AI strategy for each market" will be the mainstream approach going forward.
The outlook is that Qwen integration will increase in Japanese cars' China-market sales models from 2027 to 2028.
The era when "a car with a Japanese logo but Chinese-made AI" becomes the norm is almost here.
A. The two main differences are "market" and "level of integration."
① ChatGPT (OpenAI) has partnered with European and American manufacturers like BMW and Mercedes, centered on Western markets.
② Gemini (Google) is centered on North American manufacturers like Ford and Honda, with strong integration through Android Auto.
③ Qwen (Alibaba) is centered on Chinese manufacturers, with deep integration with Chinese domestic apps (Alipay, Trip.com, Meituan, etc.).
It's a differentiation like "the same hamburger chain but with different regional menus."
Functionally, all three are at a comparable level in core LLM performance, but "which market you operate in and which apps it integrates with" is the deciding factor in choosing.
The current state: Qwen for China, ChatGPT or Gemini for Europe and America, Gemini leading in Japan.
From 2027 onward, the choice of in-vehicle AI is expected to become just as important a decision criterion as the choice of car itself.
"An era of controlling your entire life by voice while keeping your hands on the wheel"—that is the essence of this Qwen in-vehicle AI announcement.
It marks the turning point at which Chinese EVs shifted from a battle of mere hardware to a battle of AI × software.
Industry analysts describe it as "the same historic turning point as the shift from horse-drawn carriages to automobiles a hundred years ago."
Japanese manufacturers are among the world's top tier in patent counts, yet they stand at a crossroads when it comes to "implementation speed and adaptation to the Chinese market."
On a personal level too, "the choice of in-vehicle AI" is becoming an important decision criterion when buying or driving a car from now on.
Three things you can start doing today: ① Check Chinese EV news once a month, ② Compare in-vehicle AI features when choosing your next car, ③ Try it out firsthand by renting a car during a business trip to China. These small, steady steps will become your greatest asset in developing the discernment needed for choosing a car in the AI era.
This article is a cross-post from AI Friends.
