Nvidia CEO Jensen Huang Embraces AI Inference Shift to Protect Chip Dominance
Nvidia reports massive quarterly earnings but faces investor concerns about new AI training methods. CEO Jensen Huang insists the company is well-positioned for the growing shift toward AI inference computing.
Nvidia reports over $19 billion in quarterly net income, yet investors worry about the company's future growth as AI labs change how they improve their models. Analysts press CEO Jensen Huang about how Nvidia will compete if tech companies shift toward test-time scaling, a method popularized by OpenAI's o1 model that gives AI more computing power to "think" through user prompts after they are submitted.
Huang calls test-time scaling one of the most exciting developments and a new scaling law, emphasizing that Nvidia is ready for this industry shift. This change places a much heavier focus on AI inference, which introduces new competitive challenges for Nvidia because well-funded startups like Groq and Cerebras currently build lightning-fast inference chips.
Despite these emerging threats in the inference market, Huang reassures investors that traditional pretraining scaling remains intact and continues to demand massive compute power. Industry leaders like Microsoft CEO Satya Nadella and Anthropic CEO Dario Amodei echo this sentiment, confirming that developers are still pouring massive resources into foundational model training even as they explore new post-training methods.