Nvidia and Snowflake have announced a partnership at the Snowflake Summit 2023 in Las Vegas. The collaboration aims to combine their expertise in processing foundation models for AI. Snowflake customers will now have the ability to rent cloud GPU capacity in Snowflake’s data warehouse installations and use that capacity to refine neural networks using Nvidia’s NeMo framework. Foundation models, such as large language models, are pre-trained neural networks that can be customized by customers to suit their specific needs.
This partnership reflects the growing trend of utilizing AI, particularly generative AI, as a business tool. Databricks, the developer of Apache Spark, recently acquired startup MosaicML for $1.3 billion. MosaicML offers a service for training and deploying foundation models. Snowflake will incorporate Nvidia GPU instances from its existing cloud service providers to implement this service. Snowflake also plans to extend its Snowpark developer platform with Snowpark Container Services, which will allow developers to leverage accelerated computing with Nvidia GPUs and AI software within Snowflake’s secure and governed platform.
When questioned about data protection in this partnership, Nvidia’s Manuvir Das emphasized that Snowflake has the main responsibility to ensure customer data remains within their boundaries. Das stated that Snowflake has a design construct in place to maintain data security, and the NeMo engine fits into that model. However, he also acknowledged that NeMo has a responsibility for security and that joint engineering work has been done to address this.
This collaboration follows Nvidia’s recent partnership with ServiceNow to use NeMo with ServiceNow’s customers in IT services. While the Snowflake partnership is more general purpose, the ServiceNow partnership focuses on the independent software vendor (ISV) model. ServiceNow uses the NeMo code to train customer models specific to each of their customers, providing personalized responses for their IT work and trouble tickets.
Nvidia CEO Jensen Huang has emphasized the importance of software as a growth opportunity for the company. NeMo is part of Nvidia’s enterprise software stack, which the company is promoting through partnerships with cloud providers. Nvidia’s software business currently generates hundreds of millions of dollars in revenue and is expected to continue growing.
In conclusion, the partnership between Nvidia and Snowflake aims to enhance the processing of foundation models for AI. This collaboration reflects the increasing use of AI as a business tool and follows Nvidia’s strategy to focus on software as a growth opportunity. The partnership with Snowflake is part of Nvidia’s larger enterprise software stack, which includes partnerships with cloud providers.