Data here, model there, Virtzilla’s usual abstraction principles in between
VMware has teased looming announcements outlining how it will enable generative AI to run on it stack, drawing on resources housed across multiple clouds.
Virtzilla’s hints appeared in its July 2023 multicloud briefing – a video event staged late last week. During the event, VMware president Sumit Dhawan outlined a strategy to ensure customers can run large language models efficiently in a VMware environment, and to enable fast implementation of the technology.
Much of the strategy outlined in the video is core VMware doctrine – ensuring that consistent policy is applied to resources no matter where they reside across multiple clouds. For AI, that means making it possible to use data from multiple sources securely in a model.
Veep for Cross-Cloud Services Vittorio Viarengo described VMware’s intention to deliver „choice of where the data comes from across clouds, choice of where you run your model and your application with high efficiency, but with the control and privacy that enterprise customers require.
