ZTE and ByteDance Push System AI Phone

A single voice command can order lunch—the Doubao AI phone is real.

ZTE revealed on March 31 that it is deepening its collaboration with ByteDance to accelerate the development and commercialization of a new generation of Doubao AI smartphones. The move builds on a pipeline of progress already in play: in December 2025, ZTE’s Nubia sub-brand launched a technical preview of the Nubia M153 Doubao AI phone assistant, priced at RMB 3,499. The prototype demonstrated an ambitious, system-level integration where Doubao ships with deep OS permissions to understand and execute cross‑app commands via natural language—think telling the phone “I’m hungry” and having it open a delivery app and place an order without manual app switching. ZTE executive Bai Gang described the experience as “autonomous driving for smartphones,” a phrase that underscores a shift from humans operating phones to AI operating phones for you. The engineering prototype sold out quickly, signaling appetite for a new class of AI-native devices.

Mandarin-language reporting indicates ByteDance’s strategy here is ecosystem-driven rather than hardware-centric. The company has said it has no plans to build its own hardware; instead, it aims to weave its large AI models deeply into partner operating systems. In this arrangement, ZTE supplies the underlying system adaptation and hardware-level optimization to unlock this AI-native experience. The collaboration signals a practical blueprint for how China’s fastest-moving AI capabilities may arrive on consumer devices: not through a standalone app, but through a deeply embedded, cross-app intelligence that can orchestrate actions across apps with natural language.

For observers, the project is a reminder of how the country’s manufacturing and tech ecosystems are converging around AI-enabled products. Doubao’s system-level integration demands hardware capable of sustaining on-device inference, rapid memory access, and efficient cross-application orchestration. The M153 preview price point and the rapid sell-out hint at demand for mid-range hardware capable of supporting sophisticated AI workflows, a sweet spot for domestic ODMs and component suppliers alike. If the model can run smoothly without draining the battery or triggering thermal throttling, it could reframe expectations for what constitutes “value” in a smartphone at RMB 3,500 and up.

Two practitioner tensions to watch emerge from this alliance. First, the hardware-software co-design dynamic will determine success or failure. ZTE’s role in hardware-level optimization is critical because system-level AI orchestration—cross-app commands, real-time captioning, social posting across apps—requires bandwidth, memory, and specialized AI accelerators. The more aggressive the AI load, the more crucial efficient silicon and power management become. Second, the ecosystem velocity versus risk calculus is real. ByteDance is betting on its models’ ubiquity within a partner OS, but ecosystem success depends on how seamlessly Doubao can be kept up-to-date across devices and apps, and how data flows are governed in practice. ByteDance’s stance—embed models into partner systems rather than releasing standalone hardware—reduces capital outlays andaltogether shifts risk to device-makers and app ecosystems. That means OEMs like ZTE will shoulder more responsibility for performance, privacy, and long-tail updates.

The development also reflects a broader Chinese manufacturing and tech strategy: advanced AI capabilities are being tethered to concrete hardware platforms through tightly knit OEM-software partnerships. If Doubao proves durable on the front end, it could steer demand toward domestic AI accelerators, memory solutions, and imaging pipelines that optimize on-device inference and multi-app orchestration. On the global stage, this model—AI-native software running through tightly integrated hardware—could pressure other smartphone makers to pursue similar ecosystem partnerships rather than trying to own every layer.

In the near term, the litmus test will be user experience at scale: can Doubao-driven commands reliably perform multi-step tasks across apps without glitches, and can the system maintain privacy and safety as it gains more control over device actions? If the answer is yes, China’s next wave of AI-powered handsets could redefine what a smartphone does for you—rather than what you do on it.

Sources

ZTE Teams Up with ByteDance to Develop Next-Gen Doubao AI Phone