Keep pulling the thread on Jensen Huang.
NVIDIA has approximately $100 billion in purchase commitments for its supply chain.
Some estimates place NVIDIA's purchase commitments as high as $250 billion.
NVIDIA has the supply chain capacity to support a trillion dollars in scale over the next several years.
NVIDIA achieves 10x to 100x performance improvements through algorithmic advances rather than just transistor improvements.
NVIDIA's transition from the Hopper to the Blackwell architecture yielded a 50x efficiency gain.
Jensen Huang predicts that the number of AI agents and software tool users will grow exponentially.
Jensen Huang argues that restricting chip exports to China is actively harmful to the United States.
Jensen Huang predicts that China will build its own competing AI ecosystem regardless of US export controls, which will allow them to develop scale and optimize future AI models differently than the American tech stack.
Jensen Huang believes the scaling of general-purpose computing has largely run its course.
Domain-specific acceleration provides 10x to 100x performance improvements, compared to the 25% annual improvement from Moore's Law.
NVIDIA maintains its 30-year relationship with TSMC based on trust without a formal contract.
Google trains its Gemini models on its custom TPU hardware.