NVIDIA has introduced Blackwell Ultra, a new artificial intelligence chip it says is designed to accelerate both training and inference for large language models. The company said the processor delivers about 2.5 times the performance of its predecessor, positioning it as a major upgrade for AI infrastructure.

The announcement underscores the fierce competition in the semiconductor market as firms race to supply the computing power needed for increasingly demanding AI systems. NVIDIA has dominated much of the AI chip boom, and the latest release appears aimed at reinforcing that lead as rivals push to close the gap.

According to the company, Blackwell Ultra is intended for data centers and other high-performance AI deployments where speed and efficiency are critical. If the claims hold up in real-world use, the chip could help reduce the time and cost required to run advanced AI models at scale.

The launch also highlights how quickly the AI hardware market is evolving, with chipmakers under pressure to deliver more performance without dramatically increasing power consumption. For customers building next-generation AI products, the new chip could become a key benchmark in the months ahead.