AI: Amazon and Qualcomm Partner on Chips, with a $4 Billion Option

The collaboration targets several generations of components intended for AI inference, that is, the execution of already trained models. The two groups also plan to work on very high-speed optical connectivity technologies to support the increasing computational demands in data centers.

According to a regulatory document filed by Qualcomm with the Securities and Exchange Commission, Amazon may acquire up to 25 million Qualcomm shares at an exercise price of $161.26 per share. At this level, the warrant represents a potential value of just over $4 billion if all rights are exercised.

This right does not, however, equate to an immediate investment of 4 billion dollars. Its acquisition is linked to the achievement of several commercial objectives within the framework of the relationship between the two companies, which may involve tens of billions of dollars in business.

On Wall Street, the announcement supported Qualcomm’s stock, which rose significantly on Tuesday. The group has been seeking for several years to reduce its dependence on the smartphone market by accelerating in data centers and advanced semiconductors, server processors, and AI.

Inference chips at the heart of the AI battle

The inference accelerator market is becoming one of the main battlegrounds in the semiconductor industry. While Nvidia largely dominates the training and execution infrastructures for models, major cloud groups are increasing partnerships and in-house designs to reduce costs and secure their capabilities.

Qualcomm aims to position itself in this growth phase thanks to its experience in low-power processors and a new range designed for data centers. The company is particularly focusing on energy efficiency, which has become a central issue for operators facing the increasing electricity consumption of AI infrastructures.

For Amazon, the agreement further expands AWS’s hardware ecosystem. The group is already developing its own chips for the cloud and artificial intelligence, while relying on external suppliers to meet demand that continues to grow rapidly.