Amazon cloud business is developing its own computer chips for the projects based on Artificial Intelligence.
Amazon Web Service revealed yesterday that at its AWS Reinvent user conference in the Las Vegas that its new Inferentia chips which will provide the AI researchers with the “high performance at low cost”. It is the latest example of the giant provider of the cloud services building next-gen processors.
Among the other provider of the public cloud services, Amazon is also following the Google into the chip market. Google announced its first Tensor Processing or TPU in the year 2016. Alibaba a public cloud provider that is the popular company in China has also announced its own AI chip.
AWS is by far the leader in the public cloud infrastructure services, which companies can rely remotely upon run the software and store the data. IBM, Google, and Microsoft are competing with the AWS for the business as the company moves their workloads from the traditional data centers to the cloud.
The Inferentia chips will become available by the end of 2019. Just like the other AWS services, customers will be able to pay based on how much they merely use.
There are mostly two common phrases in AI, training models by feeding them with lots of data, and then simply showing them the new data that they simply use to run the predictions. After 2016 Google has also introduced a new TPU chip which will compete with the Nvidia for the training AI models. Inferentia is focused only on inference as of now.
Amazon said that some inference workloads require an entire graphics processing unit, which is expensive. “Solving this challenge at low cost requires a dedicated inference chip,” the company said.
AWS revealed that the customers will be abel to use the Inferentia with the TensorFlow AI software, which is created by the Google, as well as some of the other AI frameworks like the ONNX and PyTorch format for converting models.
