Nvidia is set to unveil a new inference-focused AI chip at its annual GTC conference, according to the Financial Times, as the company moves to defend market share against a growing field of challengers targeting the inference segment of the AI compute market.
Inference — the process of running a trained AI model to generate outputs — has become an increasingly contested battleground as AI deployment scales. Unlike the training phase, where Nvidia's GPU dominance has been near-total, inference workloads are more cost-sensitive and better suited to a wider range of chip architectures, giving rivals an opening.
Beyond the inference chip, CNBC reports that Huang will use GTC to detail new CPU products specialised for agentic AI, tasks where models autonomously execute multi-step processes. Both Nvidia and AMD are seeing strong CPU demand in this context. The move signals that Nvidia is positioning itself across a broader stack of AI infrastructure rather than remaining anchored to GPU sales alone.
GTC has historically served as Nvidia's primary product showcase. This year's edition arrives as the company faces mounting pressure from custom silicon efforts at hyperscalers including Google and Amazon, as well as dedicated AI chip start-ups. Nvidia's market capitalisation remains among the largest globally, but its shares have come under pressure alongside broader tech sector volatility in 2026.
