Intel’s attempting to break into the AI market once more, but this time avoiding Nvidia’s dominance in training by going for inference
Intel is making another attempt to break into the AI chip market, shifting strategy to avoid direct competition with Nvidia’s dominance in AI training. Under former leadership, Intel’s Gaudi series of GPUs failed to gain traction. The company is now launching Crescent Island, a new chip focused on AI inference.
According to Financial Times reporting, Kevork Kechichian, Intel’s general manager of the data centre group, outlined this strategic pivot at Computex. By targeting inference—the computationally lighter task of running trained AI models—Intel aims to compete in a less saturated segment where Nvidia’s dominance is less overwhelming than in training workloads.
The shift represents Intel’s acknowledgment that competing directly with Nvidia on training is currently untenable. Inference is a substantial market opportunity, as deployment of trained models is critical infrastructure for enterprises and cloud providers. This approach leverages Intel’s manufacturing expertise while avoiding direct competition in training, where Nvidia’s first-mover advantage and software ecosystem have proven nearly insurmountable.