The industry barriers built by traditional smartphone manufacturers through stacking specs and competing on supply chains are rapidly crumbling under the impact of AI upstarts like Jieyue Xingchen and ByteDance. In the next decade of the smartphone industry, the power is shifting from hardware controllers to intelligent agent operators. Whoever completes this transformation first will have the opportunity to redraw the industry map.

Original Content by AI New Tech Team, Xinshang

Since the rapid rise of AI, smartphones have rarely been the center of attention in the tech world. Even Apple has scaled back its smartphone ambitions compared to its heyday. The entire industry has gradually settled into a state of calm, with everyone accustomed to making incremental improvements within existing frameworks.

However, the Chinese smartphone industry has recently undergone a subtle yet profound transformation.

On one side, Jieyue Xingchen just launched the world's first large model-native intelligent agent smartphone, the STEPX Neo. On the other, ZTE Nubia announced it will release a mass-produced AI agent smartphone at WAIC, while Honor's Robot Phone is also poised for launch.

▲Photo/Source: Internet

At first glance, this appears to be just another round of manufacturers hyping new concepts. But a deeper look at the players involved and their strategies reveals that things are far more complex.

Why Have Users Remained Unimpressed After Three Years of AI Smartphone Hype? 

As early as 2023, Qualcomm's CEO introduced the concept of AI smartphones, prompting mainstream manufacturers to quickly follow suit and spark a marketing frenzy.

Huawei, vivo, OPPO, and Honor nearly all integrated self-developed AI agents, with increasingly grandiose presentations. But when these devices reached users, the experience felt underwhelming. The so-called AI enhancements amounted to slightly better photo editing, faster voice transcription, and marginally more conversational voice assistants. Ultimately, it was just a more proactive voice assistant paired with an auto-editing photo album.

▲Photo/Source: Internet

In other words, most current AI smartphones simply overlay AI capabilities onto traditional operating systems. The OS remains unchanged, with apps operating in silos, merely adding an AI veneer. Users still need to manually switch between apps for anything moderately complex.

Jieyue Xingchen's chairman, Yin Qi, offered a precise metaphor: "Opening a door for agents in an old system makes them permanent visitors. Building a house for agents makes them native inhabitants."

For AI smartphones to escape the quagmire of concept hype, they must evolve from "OS+AI" to "AI-native OS."

True Transformation Begins at the Operating System Level

The core innovation of this wave of agent smartphones lies precisely in rebuilding from the OS level.

Take Jieyue Xingchen's Step AOS as an example. Instead of simply adding AI features to a traditional OS, it reconstructs three layers of infrastructure—computing, data, and application services—breaking down hardware capabilities into four atomic units (communication, applications, files, and system) for agents to freely combine.

The difference is immediately apparent in real-world use. When a user issues a complex command like "Plan a weekend trip, book tickets, reserve a hotel, and find a local restaurant along the way," Jieyue Amoo can automatically coordinate Ctrip, Didi, Meituan, and Alipay to complete the entire process at once, eliminating the need for users to manually switch between apps.

Nubia has taken a different approach—partnering with ByteDance, with Nubia handling hardware and ByteDance managing software. The short-lived Doubao Phone emerged from this collaboration, and Nubia will release multiple "Doubao AI Agent Smartphones" at this year's WAIC.

▲Photo/Source: Internet

Honor has introduced AgenticOS, declaring that terminals will evolve from "application containers" to "agent stages."

These three distinct paths ultimately converge. The second half of the AI smartphone race will no longer be about cramming more AI features into existing systems but about reconstructing a new OS for agents at the foundational level.

From Blockade to Embrace: What Changed?

At the STEPX Neo launch event, Jieyue Xingchen's initial ecosystem partners included Alipay, Meituan, Gaode, Didi, JD.com, Baidu, Weibo, WPS, and others—covering nearly all major daily applications.

▲Photo/Source: Internet

Yet just last year, the Doubao Phone faced collective blocking by major apps after briefly gaining popularity. Platforms cited user privacy protection, but industry insiders knew the real issue: agents bypassed apps by simulating touches, undermining platforms' traffic control and commercial models.

In just one year, attitudes have reversed. Internet giants now actively embrace agent smartphones, preferring to secure their position in new interaction paradigms rather than be bypassed.

Meanwhile, policy is paving the way. In June, the State Administration for Market Regulation explicitly stated at a news conference on agent interconnection standardization that inconsistent interfaces and protocols among manufacturers have created "agent silos," severely hindering large-scale collaborative applications. The state is now promoting unified standards, meaning cross-APP operations are moving from gray areas to compliance.

Three Major Challenges Remain Before the Landscape Solidifies

While policies and internet giants are aligning, agent smartphones still face significant hurdles to maturity.

First is security and privacy. High privileges entail high risks, especially when agents operate across apps. Abuse of permissions or system attacks could have far more severe consequences than isolated data breaches. Payment and other sensitive operations will likely require manual confirmation for the foreseeable future. Jieyue's proposed four-dimensional security framework—"trustworthy, visible, controllable, reversible"—remains untested in the market.

Second is profit distribution. The Doubao Phone's backlash last year stemmed from agents accessing services without app consent, directly challenging platforms' commercial models. While apps now voluntarily open permissions, questions about traffic accounting, commission splits, and user ownership remain unresolved. Strategic willingness can initiate cooperation, but sustainable operations depend on profit mechanisms.

Third, the current core players in the agent smartphone race are unproven. Jieyue Xingchen is just three years old, ByteDance has never independently built a smartphone, and ZTE Nubia has long been relegated to "others" in the traditional market. While they may excel in AI and systems, their ability to deliver satisfying consumer experiences remains an open question.

Of course, every terminal revolution has been led not by incumbents but by newcomers attacking from the margins. This holds true today.

The agent smartphone race may appear to be about product innovation, but at its core, it's a reallocation of industrial power. Whoever first defines the next-generation OS will control the high ground of the AI terminal ecosystem.

Yin Qi predicts: Future agent smartphones may no longer be phones at all. This vision may seem distant, but the direction is clear—smartphones will evolve from tools into user agents. Whoever completes this transformation first will reshape the terminal industry's landscape in the next decade.

For traditional smartphone giants, the clock may be ticking faster than they realize.

-END-