In December 2025’s AI track, the smartphone field has seen another heavyweight move—ByteDance has jumped out of the conventional play of “self-developed phones,” directly releasing the technical preview of the Doubao Phone Assistant and jointly launching the engineering prototype nubia M153 with ZTE. This model, priced at 3,499 yuan (approximately USD 490) and aimed at developers and tech enthusiasts, not only showcases AI’s ability to reconstruct the phone experience, but also reveals Doubao’s clear ambition on the collaborative track of “phone manufacturers + large-model vendors.” While the industry is still competing on hardware parameters, Doubao’s “light-asset entry” may be rewriting the competitive rules of AI phones.
Not making phones, only making the “brain of the phone”
Amid manufacturers like vivo, OPPO, and Huawei clustering to launch their own AI phones, ByteDance has chosen, for the time being, not to enter competition on the device side, focusing instead on an “AI assistant software at the operating-system layer.”
Behind this positioning is a judgment about the core logic of AI phones: what users truly need is not a “phone labeled with AI,” but an “intelligent collaborative capability” that is deeply integrated into daily life and automatically solves problems. The core of the Doubao Phone Assistant relies on ByteDance’s large-model technology to co-build a collaborative ecosystem of “software + hardware” with phone manufacturers: it bears neither the high costs of hardware R&D nor inventory risks, yet can quickly open up the system underpinnings through manufacturer authorization, allowing AI capabilities to truly land at the operational layer.
At present, the first batch of experience devices is on the nubia M153 in cooperation with ZTE; 30,000 demo test units went online in the first batch and sold out instantly, with some resold on secondary marketplaces at high premiums. ZTE Corporation’s shares hit the daily limit up, sparking heated market discussion. Although this engineering device was released only in small quantities, it has already sent a key signal: what Doubao aims to make is not “another phone,” but an “AI brain” that can adapt to multiple brands—later, as negotiations with more manufacturers land, ordinary users will be able to experience cross-app collaborative AI services on their existing phones without changing brands.
Can AI really “use the phone for you”? Field-tested cross-scenario automatic collaboration
In the demo video released by Doubao, the “proactive service capability” of the AI assistant breaks, for the first time, the “information silos” between mobile apps: with one sentence—“Help me book tomorrow’s high-speed rail to Shanghai and check hotel logistics”—the assistant, with user authorization, can automatically jump to 12306 to check and book tickets, simultaneously open a hotel app to filter nearby listings, and even link with a shopping app to follow up on the logistics progress of a previously ordered suitcase—the entire process requires no manual app switching by the user, completely getting rid of the hassle of “multi-step operations.”
Similar efficiency gains also appear in office scenarios: when batch-downloading email attachments, the AI, with user authorization, can automatically recognize file types and store them by category; when handling cross-software data, there is no need for manual copy-and-paste—the assistant can directly connect documents, spreadsheets, and mind-mapping tools. This “cross-app collaboration” capability is precisely the core difference between the Doubao Phone Assistant and traditional voice assistants—the former is a “partner that proactively solves problems,” while the latter is more like a “tool that passively executes commands.”
However, the official messaging is also candid: constrained by uncertainties in large-model technology, demo scenarios cannot be reproduced 100% of the time, and the current product’s usability still falls short of expectations. This “no exaggeration, no evasion” phrasing, on the contrary, makes the sincerity of commercial rollout feel more genuine.
AI phones enter the era of the “collaboration war”; going it alone is already the past
Doubao’s move is not an isolated incident, but a microcosm of the entire AI phone track. As seen in Google Gemini’s exploratory cooperation with Samsung, the current leading players have gradually realized that the key to victory in AI phones does not lie in “whose hardware is stronger,” but in “how deeply large models are fused with the system.”
On the one hand, phone manufacturers possess hardware ecosystems and user bases but lack the continuous iteration capability of large models; on the other hand, large-model vendors hold technological advantages but need hardware carriers for scenario landing—the collaboration between the two happens to fill each other’s gaps. According to IDC’s forecast, by 2027 shipments of the new generation of AI phones in China will reach 150 million units, with a market share exceeding 50%, and the cooperation model of “large models + phone manufacturers” is precisely the core lever to pry open this market.
For Doubao, this cooperation with ZTE is only a starting point; future landings with more phone manufacturers are even more worth anticipating. This “ecosystem cooperation” model can both rapidly expand the coverage of AI capabilities and, through user feedback from different brands, feed back into large-model optimization, forming a positive cycle of “scenarios – data – technology.”
What should consumers expect? Pragmatism first—no need to “pay for a concept”
For ordinary users, the value of the Doubao Phone Assistant may lie in “not forcing you to change phones for the sake of an AI concept”—in the future, whether holding a Huawei, Xiaomi, or OPPO device, it may be possible to access similar services through system upgrades, without replacing the entire ecosystem device for a single AI function.
For developers and tech enthusiasts, the 3,499-yuan (approximately USD 490) nubia M153 engineering device offers a chance to “taste it early”: not only can one field-test the real experience of cross-app collaboration, but also participate in the iterative feedback of the AI assistant. The functions of current AI phones are still in the transition from icing-on-the-cake to indispensable, and real user feedback is the key to pushing the technology toward maturity.
Of course, doubts remain: when AI intervenes deeply in phone operations, how will privacy data be protected? Can the experience of cross-brand adaptation be unified? The answers to these questions may only be revealed after Doubao’s cooperation with more manufacturers is implemented. What is certain, however, is that Doubao’s strategy of “not making phones, only making the brain” has already provided a new idea for the commercialization of AI phones—rather than chasing hardware hotspots, focusing on “collaborative innovation” that targets real user needs may be a more sustainable way to break the deadlock.
[Disclaimer]: The above content reflects analysis of publicly available information, expert insights, and BCC research. It does not constitute investment advice. BCC is not responsible for any losses resulting from reliance on the views expressed herein. Investors should exercise caution.