Forget AMD: Amazon Declares War on Nvidia by Selling Its Own AI Chips - 24/7 Wall St.

As the artificial intelligence boom accelerates, global investment in AI infrastructure has reached unprecedented levels, with major cloud providers set to spend over $750 billion in 2025 alone. Nvidi... Read original →

Amazon’s move from internal use to commercializing its AI chips marks a pivotal shift in cloud providers’ quest for compute sovereignty. Technically, while Trainium and Inferentia won’t dethrone CUDA soon, they accelerate the decoupling of AI software stacks—PyTorch and TensorFlow are increasingly hardware-agnostic, eroding NVIDIA’s ecosystem moat. From a compliance standpoint, in-house silicon insulates Amazon from U.S. export controls on advanced EUV tools and reduces geopolitical supply chain exposure. Competitively, AMD and Intel may be forced to open their software ecosystems or offer deeper customization; Broadcom could leverage VMware integration to bundle AI infrastructure. Over the next 12–24 months, hyperscalers will institutionalize 'chips-as-a-service,' compelling NVIDIA to shift from hardware-margin dominance to full-stack value delivery. If Amazon cuts per-unit AI compute costs by over 30%, pricing power in the AI chip market will structurally realign.