Intel Launches Its First Cloud-Optimized AI Chips

Intel is making a strategic shift towards AI-focused chips, introducing two new Nervana Neural Network Processors: the NNP-T1000 and NNP-I1000. These are Intel's first Application-Specific Integrated Circuits (ASICs) tailored specifically for cloud-based AI applications.

The NNP-T1000 is designed for training AI models with a balanced architecture that scales from small clusters to supercomputers. In contrast, the NNP-I1000 is optimized for high-demand inference tasks.

Additionally, Intel unveiled a next-generation Movidius Vision Processing Unit, boasting an advanced computer vision architecture that promises over 10 times the inference performance with six times better efficiency compared to competitors. While these impressive claims await real-world validation, organizations utilizing Intel for visual AI projects will likely want to explore this offering.

However, patience is necessary as the Movidius chip is not expected to ship until the first half of 2020. This innovation could signify a significant advancement in AI performance, particularly for businesses not using competitors like NVIDIA. Intel has cautioned that cutting-edge AI applications may need to achieve performance increases every 3.5 months—an ambitious goal that conventional CPUs may struggle to meet. As major tech companies like Facebook and Baidu increasingly depend on Intel for AI solutions, users can anticipate practical benefits such as faster website load times and enhanced AI capabilities.

Most people like

Find AI tools in YBX