Nvidia Developing Top Secret AI Inference Chip for Next Month Debut

The new chip is expected to be unveiled at Nvidia's annual GTC developer conference and will integrate technology from the Groq chip startup.

Published on Mar. 2, 2026

Got story updates? Submit your updates here. ›

Nvidia is reportedly working on a dedicated AI inference processor that will be used by companies like OpenAI to develop faster and more efficient AI models. The new chip is expected to be launched at Nvidia's annual GTC developer conference next month and will integrate technology the company acquired from chip startup Groq Inc. in December. Inference, the process of running pre-trained AI models in production, has emerged as a key focus area in the AI industry, with Nvidia facing competition from rivals like Google, AWS, Cerebras, and SambaNova.

Why it matters

Nvidia's GPUs have long been the dominant choice for powering AI applications, but many companies have found them to be too energy-intensive, especially for autonomous AI agents that require immense computing power. The new inference chip from Nvidia aims to provide a more efficient alternative, which could help companies like OpenAI develop more powerful AI models and applications.

The details

Nvidia's new inference chip is expected to integrate technology developed by the chip startup Groq, which Nvidia acquired in a $20 billion deal in December. Groq's chips are known as 'language processing units' and are based on a novel architecture that enables them to perform inference with much lower energy usage. OpenAI, which recently received $30 billion in funding from Nvidia, is reported to be one of the earliest adopters of the new inference chip and plans to use it to power its Codex programming tool, a rival to Anthropic's Claude Code.

  • Nvidia's new inference chip is expected to be launched at the company's annual GTC developer conference in San Jose next month (April 2026).
  • Nvidia acquired Groq Inc., the chip startup whose technology will be integrated into the new inference chip, in December 2025.

The players

Nvidia

A leading manufacturer of graphics processing units (GPUs) that are widely used in AI applications.

OpenAI

An artificial intelligence research company that has had early access to Nvidia's new inference chip and will become one of its earliest adopters.

Groq Inc.

A chip startup that Nvidia acquired in a $20 billion deal in December 2025, and whose technology will be integrated into Nvidia's new inference chip.

Cerebras Systems Inc.

A startup that develops specialized inference chips that compete with Nvidia's GPUs.

SambaNova Systems Inc.

A startup that develops dedicated inference chips that compete with Nvidia's GPUs.

Got photos? Submit your photos here. ›

What’s next

Nvidia is expected to officially launch its new inference chip at its annual GTC developer conference in San Jose next month (April 2026).

The takeaway

Nvidia's development of a dedicated AI inference chip signals the company's recognition of the growing importance of efficient inference processing in the AI industry, as companies seek to power increasingly powerful and autonomous AI applications while reducing energy consumption and costs.