At the World AI Conference in Shanghai this week, ZTE's Nubia brand unveiled the NaviX Ultra, a device that marks a radical departure from traditional smartphones. Marketed as the world's first agentic AI smartphone, the NaviX Ultra sold out its initial run of 30,000 units within hours, and prices on the secondary market quickly doubled. The device runs ByteDance's Doubao AI agent and can be activated either by voice or a dedicated hardware button, available in four colors at a launch price of 3,499 yuan ($516).
Key Facts
- ZTE NaviX Ultra debuted at WAIC 2025 as the first agentic AI smartphone.
- Powered by ByteDance's Doubao AI agent, it automates tasks across apps.
- Initial 30,000 units sold out in hours; resale prices doubled.
- StepFun and Honor also presented AI agent-based phones at the same conference.
- Chinese smartphone shipments have fallen for five consecutive quarters due to memory crisis and demand drop.
- IDC predicts over half of China's smartphone market will be AI-driven by 2026.
- Apple received approval to roll out Apple Intelligence in China via partnerships with Alibaba and Baidu.
Agentic AI: A New Paradigm
The core idea behind the NaviX Ultra and similar devices is the integration of an agentic layer into the operating system. Unlike conventional AI smartphones that bolt on isolated features—such as photo editing or voice assistants—the agentic approach allows the AI to autonomously execute complex tasks across multiple applications. For instance, a user could ask the Doubao agent to book a flight, check calendar availability, and send confirmation emails, all in one seamless interaction. Nubia chief Ni Fei criticized existing AI phones, stating that stacking AI functions on an existing system actually makes the experience more cumbersome. By embedding the agent into the OS, the phone becomes a proactive assistant rather than a passive tool.
ByteDance's Doubao is not the only agent making strides. StepFun introduced a device running a proprietary operating system with a built-in agent called Amoo, while Honor—spun off from Huawei—is co-developing an AI agent with Alibaba that will ship on new handsets later this year. Each implementation follows the same philosophy: let the AI handle cross-app workflows autonomously, reducing user friction and unlocking new possibilities for productivity.
Market Context and Competition
The timing of these launches is no accident. China's smartphone market has been under severe pressure. Shipments have declined for five consecutive quarters, driven by the global memory crisis that pushed component costs higher and consumer demand lower. IDC projects that the global smartphone market will experience its steepest annual decline on record in 2026. Chinese manufacturers, many of which rely on budget devices with razor-thin margins, are hit hardest. AI smartphones represent a potential escape route—a way to command higher prices and reignite consumer interest. IDC analyst Arthur Guo noted that more than half of China's smartphone market could be dominated by AI devices this year.
Competition with Apple adds another layer of urgency. Apple recently received Beijing's approval to launch Apple Intelligence in China through partnerships with Alibaba and Baidu. That gives Apple a backend for its AI features, but Chinese rivals argue they have a lead in hardware integration. Ni Fei asserted on Weibo in June that in terms of AI smart devices, they are ahead of Apple. The AI boom that is pressuring cheap smartphones is simultaneously creating the argument for a whole new category of phones—one where the device's intelligence becomes its primary selling point.
Beyond ZTE, StepFun, and Honor, other Chinese tech giants are investing heavily in on-device AI. Xiaomi and Oppo have also hinted at integrating large language models into their operating systems. The race to deliver a truly agentic experience is accelerating, with each company betting that consumers will pay a premium for a phone that anticipates needs and executes complex tasks without manual intervention. However, the fundamental question remains: does a smarter assistant provide enough value to drive a replacement cycle in a saturated market?
Technological Underpinnings and Challenges
Building an agentic AI smartphone requires more than just a chatbot interface. It demands tight integration between hardware, operating system, and cloud-based AI models. ByteDance's Doubao leverages the company's extensive work in natural language processing and recommendation algorithms, trained on massive datasets from its social media platforms. The agent needs to understand context, maintain state across apps, and respect user privacy and security. ZTE and its partners have had to redesign the system architecture to allow the AI to access app functions through APIs and intent-based frameworks, essentially turning the phone into a platform for AI-driven automation.
One major challenge is ensuring the agent's actions are reliable and predictable. A mistake in booking a flight or sending a message could erode user trust. Chinese regulators are also paying close attention, as autonomous AI agents raise questions about data control and decision accountability. The government has been supportive of AI innovation but has also enacted strict data privacy laws. Balancing capability with compliance will be crucial for widespread adoption.
Another challenge is battery life and processing power. Running a large language model locally for real-time tasks is energy-intensive. Most current agentic phones rely on a hybrid approach: simple tasks are handled on-device for speed and privacy, while complex reasoning is offloaded to the cloud. This requires seamless connectivity and robust edge computing. Qualcomm and MediaTek are already developing chipsets optimized for AI agents, promising dedicated neural processing units that reduce power consumption. As these chips become standard, the performance gap between agentic and traditional smartphones will narrow.
Outlook for the Industry
The success of the NaviX Ultra's initial run suggests there is pent-up demand for a smarter, more autonomous phone. But whether the second and third batches will see similar demand depends on user experience and software maturity. The agentic layer must be intuitive enough for non-technical users while remaining customizable for power users. Honor's collaboration with Alibaba aims to leverage Alibaba's cloud infrastructure and consumer ecosystem, potentially giving it an edge in integrating services like shopping, travel, and finance. StepFun, a smaller player, focuses on niche productivity applications such as automated meeting scheduling and document management. These different strategies reflect the broader uncertainty about which use cases will resonate most.
Meanwhile, the traditional smartphone market continues to struggle. Average selling prices are rising due to inflation and component shortages, but unit sales are flat or declining. AI agents could justify higher prices if they deliver measurable time savings and convenience. Early adopters are likely to be tech enthusiasts and business professionals, but mass adoption will require killer app scenarios. For example, a student using an agent to compile research notes, a traveler relying on it to handle itinerary changes, or a parent using it to order groceries and schedule childcare—these everyday use cases could eventually make the agentic phone indispensable.
The competitive landscape is also evolving beyond China. Samsung is rumored to be developing an agentic AI assistant for its Galaxy line, while Google continues to improve its own AI capabilities for Pixel devices. However, China's aggressive pace—driven by a combination of government AI initiatives, fierce domestic competition, and a mature mobile ecosystem—positions it as a laboratory for the next generation of smartphones. The NaviX Ultra is not just a product; it is a statement that the smartphone form factor is being reinvented around AI. As Ni Fei put it, phones that merely add AI features are a dead end. The future lies in phones that think and act like an agent.