Key Takeaways

  • ZTE introduced the NaviX Ultra at WAIC 2026 with an integrated ByteDance Doubao agent
  • StepFun and Honor also revealed AI-native devices as China's market turns toward agentic operating systems
  • IDC projects AI smartphones will represent more than half of China's shipments in 2026

China's smartphone industry is searching for a way out of a prolonged slump, and this week's launches at the World AI Conference in Shanghai suggest manufacturers believe the answer is an operating system rebuilt around autonomous agents. ZTE's unveiling of the NaviX Ultra, positioned as the first agentic AI smartphone, set the tone. StepFun and Honor followed closely with their own interpretations of what an AI-native device should look like.

ZTE's approach is the most concrete so far. The NaviX Ultra, under the company's Nubia brand, runs ByteDance's Doubao agent with voice activation or a hardware trigger. At a prototype price of ¥3,499 ($516), the first 30,000 units sold out in hours. A resale spike quickly doubled the price, reflecting early adopter curiosity about agent-driven mobile workflows.

Others are pushing in the same direction. StepFun's upcoming device features a proprietary operating system with a native agent named Amoo. Honor, spun off from Huawei, is demoing an AI agent co-developed with Alibaba that will ship later this year. Their shared idea is straightforward, although not simple to execute. Instead of placing isolated AI tools inside a traditional Android-style interface, the operating system becomes the orchestration layer, allowing the agent to complete tasks autonomously across apps. The Nubia president noted that many competitors simply stack AI functions on top of existing systems, making the user experience more cumbersome.

What makes this moment especially interesting is the macro backdrop. China's smartphone shipments have declined for five consecutive quarters as the memory crisis raised component prices and weakened demand. IDC expects 2026 to bring the steepest annual global smartphone decline on record. That creates pressure, particularly for vendors with thin margins. According to an IDC analyst, AI devices could represent more than half of China's smartphone market by the end of this year. A rapid mix shift like that would be unusual in mature hardware categories, although it fits a pattern seen in other consumer tech transitions noted by analysts at Gartner, who describe AI-centric device cycles as accelerators during stagnant periods.

Historically, the industry has relied on camera upgrades or form factor changes to stimulate replacement cycles. This time, manufacturers are betting that users will pay for autonomy: phones that plan travel, edit images, and take actions across apps without being asked to manual tap through menus. Whether that pitch convinces mainstream buyers remains uncertain. Some consumers still hold onto devices for four or five years, a trend pointed out in several studies by Canalys, suggesting that only a subset of users respond to software-driven refreshes.

Competition with Apple also looms over these launches. Beijing recently approved Apple Intelligence for the Chinese market, delivered through Alibaba and Baidu partnerships. The Nubia president has argued publicly that Chinese manufacturers are ahead of Apple in shipping AI smart devices. That said, Apple's presence still shapes expectations, especially for enterprise buyers who value ecosystem stability even more than raw features. If Apple's agentic model catches on globally later in the year, the Chinese vendors' early lead may narrow quickly.

Then there is the question of infrastructure. Agentic phones depend on consistent on-device compute, high-quality model integration, and predictable connectivity. Telecom experts at GSMA Intelligence have repeatedly mentioned that network readiness often determines real-world performance. If carriers struggle to optimize latency for agentic workflows, some of the promised benefits could stall.

Another angle worth watching is developer adoption. Agentic layers only deliver functional value if applications expose features that agents can reliably call. At the moment, most demonstrations rely on first-party apps that are tightly integrated with the vendor's operating system. Building a broader ecosystem will require tooling, documentation, and incentives. It is not clear yet whether Chinese vendors plan to open these systems widely or maintain vertical control. Enterprise app developers will likely push for standardized interfaces if agentic workflows become central to productivity.

From a business standpoint, the shift could help differentiate devices in a crowded budget segment. Many Chinese manufacturers have struggled to maintain margins on lower-cost smartphones. Embedding an AI agent at the core of the operating system provides a narrative that is different from processor speed or display quality. Whether that narrative holds depends on execution. If the agents occasionally misfire or trigger unintended tasks, users might disable them as quickly as they turned off earlier voice assistants.

On the consumer side, the pitch is convenience. On the vendor side, the motive is survival. The NaviX Ultra's fast sellout signals demand, but it may also reflect limited initial supply. StepFun and Honor are hoping that broader availability later in 2026 will clarify the market's appetite. If the AI boom phases out cheap phones, the market will determine whether buyers accept higher prices in exchange for autonomy, or if incremental intelligence fails to justify a hardware replacement.

The next few quarters will test that hypothesis. Chinese vendors are betting heavily that agentic smartphones can revive the category. Apple's entry into China's AI device segment raises the stakes even more. For now, ZTE, StepFun, and Honor have pushed the first wave of AI-native smartphones into the spotlight. Whether that momentum carries into mainstream adoption will depend on real productivity gains rather than just the promise of autonomous software.