Generative artificial intelligence today essentially involves cloud infrastructures. Nvidia and Asus think they can convince you to do without it.
Several large companies have decided to ban the use of AI on their premises. Apple and Samsung are fairly well-known examples of this, for reasons of sharing data with the cloud used by these new assistants like ChatGPT. Asus and Nvidia may have an idea to fix this problem.
A homemade ChatGPT server
AFS Appliance is the name of the product presented by Asus at Computex in partnership with Nvidia. It is neither more nor less than a server entirely dedicated to generative AI. All the hardware is therefore installed directly on the premises of the company rather than on a remote cloud infrastructure where it would not have control.
On board is the LLM Formosa, a competitor to OpenAI’s GPT 3.5. Beware of the invoice: for installation and maintenance, Asus intends to sell its service at 6000 dollars per month. The most greedy customers will be able to go up to 10,000 dollars per month to access an Nvidia DGX supercomputer.
According to Asus, the rapid development of ChatGPT and generative AI solutions made companies realize that they needed computing power. Associated with the need for confidentiality of companies for their projects, the marketing of the Asus AFS Appliance service was then all found. Banks or hospitals would be the first customers targeted by this service.
If we focus a lot on AIs open to users like ChatGPT, Midjourney or Google Bard, in the shadows, it is Nvidia that is doing well and is champion of the AI race. The giant saw its share price triple in just five months.
To follow us, we invite you to download our Android and iOS application. You can read our articles, files, and watch our latest YouTube videos.