Storj, a distributed storage vendor, recently acquired cloud computing vendor Valdi, adding GPU compute cycles on demand to its portfolio. This new offering aims to make generative AI training more accessible compared to hyperscaler options. The acquisition details were not disclosed.
According to Ray Lucchesi, founder of Silverton Consulting, purchasing compute on demand can benefit organizations looking to experiment with generative AI without the need to invest in GPUs or commit to a single provider. Lower-cost storage options could also attract more customers to the service, as moving data in and out of cloud compute can be expensive and difficult to manage.
Valdi, founded in 2022, has built a network of partners to provide access to unused GPU computing resources for high-performance workloads. Their customers can use the servers as needed without a contract and pay via credit card. Storj, on the other hand, allows the general public to resell excess storage space and offers S3-compatible object storage.
Both Storj and Valdi are working on integrating their technologies to create an alternative public cloud for compute and storage, aiming to compete with hyperscalers. Mitch Lewis, an analyst at Futurum Group, noted that Storj has successfully transitioned from a cryptocurrency project to a technology provider and storage company, with a focus on data security, privacy, cost-efficiency, and scalability.
However, Storj will need to demonstrate the performance of its technology when pushing data to GPUs for AI training, especially due to its distributed nature. Supporting technologies like GPUDirect software will be essential for faster data transfer to GPUs. Overall, Storj aims to provide worldwide access to generative AI technology while utilizing existing data center resources to reduce environmental impact.