Qualcomm bets on AI chips to lead the next generation of devices

What happened?

Qualcomm, the US semiconductor giant known for its Snapdragon chips powering most Android smartphones, has announced two new products focused on artificial intelligence and revealed it is working on more than 40 AI hardware designs, according to TechCrunch. CEO Cristiano Amon stated that the company aspires to be the chip inside devices that will replace the smartphone, betting on a future where AI is integrated into all devices. These new designs range from wearables to smart home, automotive, and industrial devices, marking a significant shift in the company's strategy.

Why is it important?

This move signals a key strategic shift. Qualcomm seeks to reduce its dependence on the smartphone market, which has shown signs of saturation: according to IDC, global smartphone shipments fell 3.2% in 2023. The company is now positioning itself in emerging markets such as wearables, smart home, automotive, and industry. The bet on edge AI is crucial: instead of relying on the cloud, devices will be able to run AI models locally, improving privacy, latency, and energy efficiency. This is especially relevant in applications like voice assistants, augmented reality, and autonomous driving, where response speed is critical. Additionally, edge AI reduces bandwidth costs and avoids dependence on stable internet connections.

Consequences and context

Historically, Qualcomm has dominated the connectivity and mobile processing segment, with approximately 40% market share in smartphone chips. However, with the advent of generative AI, demand for specialized chips (NPUs, neural processing units) has grown exponentially. Competitors like Apple with its Neural Engine, Google with its Tensor Processing Unit (TPU) in Pixel devices, and MediaTek with its APU (AI Processing Unit) already integrate AI capabilities into their chips. Qualcomm, with its portfolio of over 40 designs, aims to keep up and offer vertical solutions for different sectors. This could accelerate the adoption of AI assistants in everyday devices, from smart glasses to home robots. A concrete example is the Snapdragon X Elite, introduced in 2023, which includes an NPU capable of running large language models (LLMs) locally. Additionally, Qualcomm has collaborated with Meta to optimize Llama 2 on its chips, demonstrating its commitment to generative AI on the edge.

What readers should know

For consumers, this means upcoming devices could be smarter and more autonomous, with AI capabilities that work offline. For instance, a voice assistant could process commands without sending data to the cloud, improving privacy. For businesses, it represents an opportunity to integrate AI into existing products, such as smart surveillance systems or industrial robots. However, market fragmentation and competition from giants like NVIDIA and AMD in the AI space could limit Qualcomm's success. NVIDIA dominates the server AI chip market (with over 80% share), while AMD advances with its Instinct accelerators. Qualcomm must demonstrate that its chips offer superior performance in AI tasks without sacrificing energy efficiency, a difficult balance to achieve. Additionally, the company faces challenges in the PC market, where Intel and AMD are historical leaders. Nevertheless, Qualcomm's expertise in connectivity (5G, Wi-Fi) and low power consumption gives it an advantage in mobile and IoT devices. The company has also announced partnerships with automakers like BMW and Mercedes-Benz to integrate its Snapdragon Digital Chassis platforms, which combine connectivity, processing, and AI for software-defined vehicles. In the wearables space, the Snapdragon W5 Gen 1 already offers AI capabilities for health monitoring and smart notifications. With these moves, Qualcomm positions itself as a key player in the edge AI ecosystem, but success will depend on its ability to execute and differentiate in an increasingly competitive market.