Doubao AI Smartphone (nubia M153) Specs

According to Counterpoint’s AI 360 report, over 30% of global smartphone shipments in 2025 will support GenAI, up from 20% in 2024 and expected to reach 57% by 2029. GenAI capabilities are increasingly standard in high-end models and are expected to accelerate penetration into the mid-tier segment. This rapid AI adoption is driving significantly higher hardware demands. Current testing of the Doubao AI smartphone is based on the Snapdragon 8 Elite platform, and future development will drive high-end chips toward greater computing power, with NPUs reaching 80-100 TOPS. Running AI features significantly increases RAM and ROM requirements and drives the need for high-bandwidth memory. Additionally, AI workloads raise power consumption by 8%-12%, posing new challenges for battery technology and overall power-efficiency optimization.

Doubao’s high-privilege AI clash with app ecosystem

The Doubao AI smartphone’s core strength lies in its intent framework and cross-app orchestration, leveraging visual and semantic understanding to break user instructions into sub-tasks and execute them across multiple apps. While it performs well in single-app, clear-instruction scenarios, error rates remain high in multi-app, complex semantic tasks, with overall accuracy at roughly 50% of expected performance. Improving voice recognition, visual understanding and multi-app execution accuracy remains a key priority for refinement.

However, Doubao’s high-privilege simulated operations push against the security and commercial boundaries of the existing app ecosystem. Automated actions in apps like WeChat, Taobao, Alipay and banking apps often trigger security alerts, forced logouts, or even account bans. At the core of this conflict, super apps interpret Doubao’s simulated taps as unauthorized automation, activating strict risk controls, while its cross-app capabilities threaten their control over user traffic and commercial entry points.

For WeChat, whose core value rests on social relationships and payment infrastructure, any AI-driven automation that accesses chat histories or initiates payment actions poses a direct threat to user-privacy safeguards and financial security. For e-commerce platforms, price comparison strikes at the heart of their business model. They invest heavily in product databases, merchant systems, and recommendation algorithms to retain users within their own ecosystem. AI-driven cross-platform comparison not only diverts transactions but also undermines traffic allocation and advertising revenue systems that they have meticulously built.

Open cooperation, strategic rivalry

ByteDance and ZTE are collaborating in a model similar to the Seres-Huawei partnership – Doubao provides the core AI capabilities, while Nubia is responsible for hardware manufacturing. Doubao’s move into the smartphone space is primarily aimed at securing a system-level AI entry point. Previously, Doubao was just one app among many, constrained by smartphone OEM permission controls. By entering the smartphone domain, Doubao can be embedded directly into the system layer, becoming the default AI agent and evolving from a passive, on-demand tool into an active initiator and executor of tasks. This also enables tighter integration of ByteDance’s computing power, large language models and traffic scenarios, forming a more closed-loop AI ecosystem.

However, major smartphone OEMs with more mature in-house ecosystems are unlikely to adopt this Seres-Huawei approach. They are exploring more open and cooperative APP ecosystem frameworks. For example, OPPO has partnered with Alipay through the ‘Agent Hub Access’ framework, allowing system-level AI to work securely and compliantly with Alipay’s agents.

Looking ahead, smartphone OEMs and top apps may converge on standardized ecosystem interfaces, with system-level AI acting as the central orchestrator. With explicit user authorization, AI will invoke specific services via official, standardized APIs provided by the apps – rather than simulated taps – to execute cross-app tasks. Revenue opportunities may come from joint membership, service-fee sharing, or efficiency-driven incremental transactions. Meanwhile, control over high-value data and data security will remain a long-term area of negotiation.