Snowflake’s New Partnership with NVIDIA Grants Users the Power of AI
Snowflake and NVIDIA issued a joint press release at the beginning of the summit, in which they characterized their collaboration as “a new opportunity for enterprises. It will enable them to use their proprietary data — which can range from hundreds of terabytes to petabytes of raw and curated business information — to create and fine-tune custom LLMs [large language models] that power business-specific applications and services.”
Grabs cited the new partnership with NVIDIA when explaining the opportunities that are becoming available to Snowflake customers. “The partnership with NVIDIA is fundamentally rooted in our ability to run containers that are backed by NVIDIA-accelerated hardware, GPUs, and then also layering a whole software stack from NVIDIA. On top of that, we are also enabling our partners and also our customers that want to build out their own machine learning practice,” he said.
When using AI and ML to manage information, many users still have concerns about privacy, especially when it comes to highly sensitive data. According to Grabs, the partnership with NVIDIA not only offers new capabilities but also accounts for the need to secure the data being used. “A lot of the sensitivities that customers have today interacting with general AI — particularly from the enterprise — is around how sensitive the data is. And where am I sending my data to? If it's hosted as an LLM somewhere in the cloud, can I trust it? How much want do I want to rely on that? Wouldn't it be easier for me if I actually ran that LLM capability inside my security perimeter in Snowflake? And now, with the announcements to date, that literally has become possible, right? And we've shown how you can do that with NVIDIA.”
Data Governance, Compliance and Privacy Remain Concerns
Of course, managing voluminous amounts of data brings up issues around governance and compliance with privacy laws, such as the General Data Protection Regulation. Grabs related the stories of some customers that wanted to store their data on the Snowflake platform because of its compute power, but then wanted to make copies to be used in another cloud platform.
“As soon as you introduce a redundant copy of the data, you create all sorts of problems for yourself as an enterprise. There's governance and compliance issues that immediately pop up. So under GDPR, for example, how do you make sure that you keep track of that data so that you are compliant?” he asked.
“There's versions-of-the-truth issues that these copies may, over time, start to drift and become their own silos. All of that is a lot of work, unnecessary work. And it also introduces unnecessary, unwanted risk and complexity into a customer's data stack,” Grabs explained. “It is so much easier to actually take the compute and bring it to where the data sits, rather than the other way around, because of all these complexity issues around governance, compliance and security and privacy.”