French AI startup FlexAI will release its first solution later this year. It will allow AI developers ‘as-a-service’ and ‘pay-as-you-go’ access to various multi-cloud environments for running (AI) workloads. The startup raised 28.5 million euros in investment for this purpose.
FlexIAI has been active since late last year but is now announcing its first product for AI developers, according to TechCrunch. This solution, which should become available later this year, offers developers access to multiple multi-cloud architectures to run their own LLMs and underlying workloads as they see fit.
Like public cloud services, the new infrastructure service allows AI developers to run workloads without worrying about which GPUs, networking and storage to use to do so.
On-demand cloud access for AI infra
To this end, FlexAI offers on-demand cloud access for training AI models. The startup calls this access to ‘virtual heterogeneous compute’. FlexAI’s environment abstracts the infrastructure and associated complexity required for AI work, allowing developers to focus entirely on their AI (training) work and pay on a usage basis rather than paying based on GPU rental hours.
The startup acts as a so-called ‘aggregator of demand’ and hires the required hardware capacity from providers in the traditional way. FlexAI claims to have strong connections with GPU providers such as Intel and AMD, but can also provide GPU capabilities from Nvidia, Graphcore, or Cerebras as customers desire. These strong connections would enable FlexAI to ensure the best capacity cost for its as-a-service customers.
Financing and future plans
Before its launch, FlexAI had already secured a startup budget of 28.5 million euros through venture capital investment rounds. The startup is led by CEO Brijesh Tripathi. His experience includes working at Nvidia, Tesla, and Zoox and, more recently, as VP Intel AI at supercomputing spin-off AXG.
In the long term, FlexAI also plans to offer its own infrastructure from its data centers. To finance these, the startup plans to secure bank loans or use the services of other money lenders instead of holding additional investment rounds and issuing shares.
Also read: Nvidia CEO: Every country must build its own AI infrastructure