Cerebras Systems launched on Tuesday (27) a tool for software developers artificial intelligence (AI) that allows them to access the startup's gigantic chips to run applications, offering what it says is a much cheaper option than industry standard processors, from Nvidia.
ADVERTISING
Access to Nvidia graphics processing units (GPUs) – often through a cloud computing provider – to train and deploy large artificial intelligence models used for applications such as ChatGPT da OpenAI can be difficult to obtain and expensive to execute, a process developers call inference.
“We are delivering performance that cannot be achieved by a GPU,” said Cerebras CEO Andrew Feldman in an interview with Reuters. “We are doing it with the highest precision and we are offering it at the lowest price.”
The inference part of the AI market is expected to be fast-growing and attractive – ultimately worth tens of billions of dollars if consumers and businesses adopt AI tools.
ADVERTISING
The Sunnyvale, California-based company plans to offer several flavors of the inference product through a developer key and its cloud. The company will also sell its AI systems to customers who prefer to operate their own data centers.
Cerebras' chips – each the size of a dinner plate and called Wafer Scale Engines – avoid one of the problems with computing AI data: The data computed by large models that power AI applications typically cannot fit on a single chip and can require hundreds or thousands of chips chained together.
This means Cerebras chips can achieve faster performance, Feldman said..
ADVERTISING
It plans to charge users as little as 10 cents per million tokens, which is one way companies can measure the amount of data output from a large model.
Read also
