Qualcomm’s Cristiano Amon: Driving Terminal AI Adoption with Industry Ecosystem Collaboration

July 5 — At the 9th Gmosemiconsductor Conference’s main summit in 2025, Jerry, Chairman of Qualcomm China, delivered a keynote address. Addressing the theme of how terminal-side AI is revolutionizing the semiconductor industry and creating new opportunities, he shared Qualcomm’s insights from the terminal and mobile computing markets.

This marks Qualcomm’s ninth appearance at the Gmosemiconsductor Conference, a testament to its engagement with the industry. Jerry extended his congratulations on the successful staging of the event. He noted that over the past nine years, Qualcomm has experienced robust growth in China, largely due to the nation’s rapid economic development and the swift global rise of its terminal industry supply chain.

“Specifically in Zhangjiang, our engineering talent has doubled over the last nine years, a pace of growth that is uncommon among multinational corporations, particularly U.S. tech companies. We are deeply grateful for the significant support from the Pudong district government and various departments within the Shanghai municipal government,” Jerry stated.

AI Emerges as the New User Interface, Creating Structural Demand for the Chip Industry

Jerry pointed out that the semiconductor industry’s rapid expansion over the past decade was largely fueled by the widespread adoption of mobile internet and smart terminals. However, starting in 2024, generative AI has emerged as a pivotal new variable, fundamentally reshaping the underlying logic of the entire computing industry. This transformation is not merely about advancements in model parameters; it signifies AI’s migration from the cloud to the terminal, moving from a future that was once envisioned to a present where it is “affordable and accessible.”

In this transition, the value proposition of terminal-side AI is becoming increasingly evident. It enables rapid on-device responses to user needs, safeguards privacy by keeping data localized, and delivers more consistent and personalized intelligent experiences. Whether it’s a voice assistant on a smartphone or an intelligent cockpit in a vehicle, terminal-side AI is making smart services faster, more secure, and more intuitive.

Concurrently, a clear trend is emerging: the future of AI is not an “either/or” choice between the cloud and the edge. The capabilities of cloud-based models and the real-time responsiveness of terminals must complement and evolve together.

“This is precisely what Qualcomm emphasizes with its hybrid AI architecture – enabling dynamic task allocation between the cloud and the terminal to achieve a balance of performance, efficiency, and user experience. In essence, it means handling routine tasks quickly on your device and leveraging the cloud for complex or demanding computations,” Jerry elaborated.

Jerry highlighted that AI is rapidly becoming the new user interface. Gone are the days of manually navigating through apps and functions; intelligent AI agents are now redefining human-computer interaction through natural modalities like conversation, voice, and vision. Users no longer need to launch specific applications; they can simply express their needs, and AI agents will analyze, understand intent in real-time, orchestrate task distribution and planning, and ultimately, interact with various services for entertainment, e-commerce, navigation, and more.

Consider a business travel scenario: upon receiving a meeting invitation email, a terminal-side AI agent can automatically recognize the travel intent. By integrating with local calendar data, past travel records, and user travel preferences, it can quickly generate a preliminary itinerary, including flight times, hotel bookings, and navigation to the meeting venue. Furthermore, the agent can proactively remind the user to prepare necessary documents and even organize meeting materials in advance. During the journey, the AI can dynamically adjust departure times by referencing weather, traffic, and flight updates. Upon arrival, it can recommend nearby lunch spots based on the user’s dining preferences and the meeting location. This entire process unfolds seamlessly without manual intervention, making business travel more efficient, organized, and tailored to user habits and needs.

Jerry emphasized that AI is supplanting the “app-centric” interaction model, evolving into a truly “human-centric” intelligent gateway. This shift not only transforms the user experience but also positions terminal-side AI as an indispensable capability for the next generation of computing platforms, consequently creating new, structural demands across the entire semiconductor industry.

Hardware-Software Synergy Fuels On-Device AI Performance

According to IDC forecasts, the global generative AI market is projected to grow at a compound annual growth rate of 63.8%, with its market size expected to surpass $280 billion by 2028. Amidst this robust growth trajectory, the industry widely regards 2025 as the inaugural year for AI on terminals, with terminal-side AI rapidly becoming a key focus area for major manufacturers.

As generative AI extends from the cloud to the terminal, the entire semiconductor industry is poised for a “chip” opportunity. Industry data indicates that the global semiconductor industry is anticipated to achieve a 15% year-over-year increase in 2025, driven by escalated demand for AI and High-Performance Computing (HPC). Looking ahead, AI is poised to be a sustained growth driver for the chip sector, with projections suggesting global chip sales could exceed the trillion-dollar mark by 2030.

Jerry stated that against this backdrop, AI is accelerating its descent to the edge and terminals. This expansion not only broadens the application scope of AI but also imposes novel requirements on chip architectures, necessitating the deep integration of AI principles from the initial stages of chip design into the entire System-on-Chip (SoC).

Jerry explained that as exemplified by Qualcomm’s latest Snapdragon platforms, the company has engineered a truly heterogeneous computing system. This system deeply integrates multiple processing units, including AI-specific Neural Processing Units (NPUs), high-performance CPUs, GPUs, Image Signal Processors (ISPs), and Digital Signal Processors (DSPs). AI applications can dynamically allocate tasks across these compute units based on their specific characteristics, thereby optimizing for performance, power efficiency, and response speed for the best operational outcome.

“For instance, for spontaneous, latency-sensitive tasks, we leverage general-purpose hardware acceleration units like CPUs and GPUs to ensure rapid processing. Conversely, for scenarios requiring sustained high compute power, such as photography, video recording, and generative AI applications, the heavy lifting is handled by our power-efficient NPUs,” Jerry elaborated.

It is through this task-driven, resource-orchestrated intelligent architecture that terminal devices can truly deliver an “affordable, accessible, and superior” AI experience.

However, powerful hardware alone is insufficient. For AI to be effectively implemented on terminals, hardware and software synergy is paramount. This means Qualcomm must not only build robust heterogeneous computing capabilities at the chip level but also undertake extensive system-level optimizations at the platform level. This includes developing a comprehensive AI toolchain and software framework that covers model deployment, quantization, compilation, and runtime scheduling and tuning. This ecosystem is designed to empower developers to efficiently utilize every compute resource, ensuring that AI runs effectively, quickly, and stably on terminals, all while minimizing power consumption.

Collaborating with the Industry Ecosystem to Drive Terminal AI Adoption

Currently, China stands as the world’s largest market for smart devices and one of the most vibrant hubs for AI innovation. Jerry expressed his pleasure in observing Chinese smartphone, smart vehicle, and IoT terminal manufacturers emerging as leading forces in global terminal-side AI applications.

Qualcomm is actively partnering with the entire terminal ecosystem to foster the adoption and widespread implementation of terminal-side AI. In the smartphone sector, Qualcomm collaborates with numerous Chinese manufacturers, including Xiaomi, Honor, OPPO, and vivo, to introduce flagship devices powered by the Snapdragon 8 Gen 1 Mobile Platform, which support rich generative AI use cases. Currently, there are over 100 designs based on the Snapdragon 8 Gen 1, encompassing both currently released and in-development products.

In the PC market, industry data indicates that AI PCs constituted 17% of global PC shipments in 2024, with this proportion projected to surge to 70% by 2028. By 2028, shipments of next-generation AI PCs in the Chinese market are expected to be 60 times higher than in 2024.

Jerry noted that in line with this trend, Qualcomm is collaborating with industry partners to build a robust AI PC ecosystem. Currently, Qualcomm’s partners have launched over 80 AI PC devices equipped with the Snapdragon X Elite compute platform, and this number is expected to exceed 100 by next year.

Among various terminal form factors, automobiles are emerging as a significant vessel for terminal-side AI implementation. Vehicles possess an abundance of sensors, powerful computing platforms, and diverse interaction scenarios, all of which place higher demands on AI-driven real-time processing, reliability, and personalization. Qualcomm has partnered with several leading Chinese automotive manufacturers to accelerate the deployment of intelligent experiences. In the past three-plus years alone, numerous Chinese automotive brands have launched over 210 models based on the Snapdragon Digital Chassis. To deepen its collaboration with China’s automotive industry ecosystem, Qualcomm has hosted its Automotive Technology and Cooperation Summit in China for three consecutive years.

Jerry emphasized that from smartphones to AI PCs, and from intelligent connected vehicles to smart glasses and head-mounted displays, the applications of generative AI are rapidly expanding across all industries, seamlessly integrating into every facet of users’ lives. In response to this trend, semiconductor companies must not only provide powerful computing platforms but also collaborate closely with software, algorithm, and application ecosystem partners to build a truly open and sustainable intelligent terminal ecosystem.

Looking ahead, Jerry conveyed Qualcomm’s commitment to continuing its collaboration with industry partners to accelerate the deployment of AI on terminals, collectively embracing the “chip” growth driven by terminal-side AI, and exploring even broader future opportunities.