Tech giants develop custom AI chips to rival Nvidia

Major technology companies, including Amazon, Google, and Microsoft, are developing their own artificial intelligence processors to reduce dependency on Nvidia. While these firms remain among Nvidia’s largest customers, creating proprietary chips like Amazon’s Trainium allows them to retain more revenue within their own ecosystems.

Nvidia maintains a significant advantage through its CUDA software ecosystem and its ability to connect thousands of processors into massive computing systems. While companies like TSMC manufacture the physical chips based on designs from Nvidia, AMD, and the tech giants, the industry is shifting toward specialized AI accelerators.

Simultaneously, the application of AI is moving toward edge computing. While large-scale data centers remain essential for training massive models, the phase of inference—where models make real-time decisions—requires lower latency. For autonomous vehicles, industrial robotics, and smart infrastructure, processing data at the edge can reduce latency from hundreds of milliseconds in centralized clouds to as little as 1 to 30 milliseconds.