On April 9, it was reported that Amazon CEO Andy Jassy revealed the company is considering changing its strategy for self-developed chips, and in the future will directly sell its AI chips and corresponding rack systems to third parties.

Key Shift: From "Renting Computing Power" to "Selling Hardware"

Previously, Amazon's Trainium series AI accelerators were mainly leased to customers through AWS cloud services. This strategic adjustment means Amazon will directly challenge NVIDIA's dominance in the hardware market:

50 Billion Dollar Vision: Industry estimates suggest that if this chip division operates independently and opens up fully, its annual revenue could jump from the current $20 billion to $50 billion.

High Cost-Performance Advantage: The second-generation AI chip Trainium2 has a cost-performance ratio about 30% higher than similar products, showing strong market competitiveness.

High Demand: The upcoming Trainium3 is even more impressive, with subscription rates already close to full capacity, indicating a strong market demand for alternatives to NVIDIA.

Strategic Background: The "De-NVIDIA" Movement of Cloud Giants

Amazon's move reflects the common anxiety and ambition of top tech companies in the AI era:

Cost Reduction: By selling self-developed chips on a large scale, Amazon can spread out the high R&D costs and reduce reliance on expensive NVIDIA GPUs.

Building a Full-Stack Ecosystem: From the underlying chips, racks, to cloud platforms and upper-level large models, Amazon is trying to build a completely self-controlled AI supply chain.

Computing Power Accessibility: With a 30% cost-performance advantage, Amazon is expected to attract many small and medium-sized AI startups that are cost-sensitive.

Industry Impact: A Possible Transformation in the AI Computing Market

Currently, the global AI computing market has long been in a state of "high demand and limited supply." If Amazon (China) Investment Co., Ltd. and its parent company successfully achieve chip exports, it would provide the most important alternative computing power option besides NVIDIA, thus easing supply chain pressure and driving down overall industry costs.

Conclusion: A Computing Power Player Moving from the Backstage to the Forefront

When Amazon no longer satisfies itself with being the largest "cloud buyer," but instead becomes a "chip seller," the competition landscape of the AI hardware market will be reshaped. With the reputation accumulated by the Trainium series, Andy Jassy is leading Amazon to sprint rapidly in the trillion-dollar AI hardware race.