Nvidia, a company that sells graphics cards for computers and other devices, Wednesday announced the launch of a cloud service for developers to train artificial intelligence models.
What it means is that Nvidia plans to directly compete with Amazon Web Services, Microsoft's Azure, and Alphabet's Google Cloud Platform — to whom it now sells its graphics processing units for their cloud services. Those cloud services, in turn, provide GPU-backed virtual machine instances that developers use to run their AI workloads.
The move could lead companies that need these services to forgo the biggest cloud companies and go directly to Nvidia. Naturally, those big cloud companies that are buying GPUs from Nvidia now aren't going to look favorably on this move, and will likely try to enhance their offerings or lower their prices.
To be clear, Nvidia isn't building a whole cloud infrastructure from scratch. Instead it will rely on public cloud providers like AWS to run its service. But Nvidia will still be competing with those providers in this very particular space.
The new service will become available in public beta in the third quarter, Nvidia said in a blog post. Pricing information is not yet available.
The Nvidia service will let developers use frameworks like CNTK (from Microsoft), MXNet (promoted by Amazon) or TensorFlow (from Alphabet) for "deep learning," a type of AI that involves training artificial neural networks on lots of data and then getting them to make decisions based on that.
Intel, the biggest U.S. chipmaker, has yet to introduce a similar service but could do so in the future.
Yesterday Nvidia beat analysts' estimates for quarterly earnings per share and revenue.
Watch: 4 things propelling Nvidia