ASUS wants to give companies more control over the security of their data when using generative AI tools. To this end, the tech giant is introducing the generative AI AFS Appliance server that companies can place in their own data center environments.
The use of generative AI in enterprise environments is often done through cloud-based AI tools and solutions. Many companies are therefore concerned that business-sensitive data is uploaded to these environments, think ChatGPT, “learned” by the tools and thus made available elsewhere.
Introducing ASUS AFS Appliance
ASUS wants to respond to these fears and therefore announced the AFS Appliance, Bloomberg reports. This is a dedicated AI server that companies can install in their own data center environment, giving them control over the use of generative AI.
The AFS Appliance includes Nvidia A100 GPUs on board. The appliance is offered by the Taiwanese tech giant on a rental basis only. According to Bloomberg, the rental price is about $6,000 per month. The most expensive model of the AFS Appliance, based on the Nvidia DGX AI supercomputing platform, is likely to have a rental price of $10,000 per month.
Proprietary LLM model
Remarkably, the dedicated server runs its own LLM model called “Formosa Foundation Model,” a model developed by a subsidiary of ASUS that is said to have the same capacity as GPT 3.5.
The model is said to mainly generate texts based on traditional Chinese language structures. This indicates that the AFS Appliance is probably primarily targeting the Chinese market at this time.
It is not yet known when the ASUS AFS Appliance will become available. The tech giant first wants to have between 20 and 30 Taiwanese customers and expand internationally by the end of this year.