Snowflake pioneers customized AI applications with NVIDIA

Snowflake CEO Sridhar Ramaswamy on a conference stage next to a screen showing NVIDIA CEO Jensen Huang | Snowflake pioneers customized AI Applications with NVIDIA

Driving around San Francisco, where the Snowflake Summit 2024 is currently taking place, some things are certain, as well as admiring the smart city network and the prevalence of driverless taxis, nine in ten billboards will tell you about the power of AI and how you can leverage its benefits for your business. 

As the home of Silicon Valley, the city has attracted entrepreneurship, fresh thinking and emerging talent that is often on the lookout for ways to improve ways of working.

Fully resonating with the local atmosphere, Snowflake unveiled a bunch of developments to boost how much organizations and data analysts can get out of its offerings, including unlocking the power of AI across teams.

In his keynote, Snowflake’s Benoit Dageville, co-founder and president of product, emphasized that “after all, enterprise AI is only as good as the data foundation it is built on” while listing the top most important elements of the data journey – like “elastic on-demand compute for both CPUs and GPUs, enabling you to easily build and run any workload; access to world-class models, in a serverless version optimized for efficiency and governance at every layer; and the ability to share and access AI models and apps, not only within your organization, but also cross-cloud, cross-region and even cross board”.

Snowflake proceeded to announce a new collaboration with NVIDIA to help customers and partners build customized AI data applications in Snowflake, powered by NVIDIA AI. 

In this string of partnerships, Snowflake has adopted NVIDIA AI Enterprise software to integrate NeMo Retriever microservices into Snowflake Cortex AI, its fully managed large language model and vector search service. 

This is poised to enable organizations to connect custom models to diverse business data and deliver highly accurate responses. In addition, Snowflake Arctic, the open-grade LLM, is now fully supported with NVIDIA TensorRT-LLM software, to provide users with highly optimized performance. Arctic is also now available as an NVIDIA NIM inference microservice, allowing more developers to access Arctic’s efficient intelligence.

As enterprises look for ways to apply their data strategically and drive customization, through Snowflake’s collaboration with NVIDIA, users would be able to more rapidly create bespoke, use-case-specific AI solutions and realize the potential of enterprise AI. 

“Pairing NVIDIA’s full stack accelerated computing and software with Snowflake’s state-of-the-art AI capabilities in Cortex AI is game-changing,” said Sridhar Ramaswamy, Snowflake’s CEO. “Together, we are unlocking a new era of AI where customers from every industry and every skill level can build custom AI applications on their enterprise data with ease, efficiency and trust.”

Snowflake equips developers with building skills in Snowflake Cortex AI

Snowflake also unveiled new tools to accelerate how developers build enterprise-grade pipelines, models and applications with their data. 

In this way, the company is furthering its mission of eliminating complexity for customers with new developer tools and native integrations that speed up development and empower them to ship more advanced products in the AI Data Cloud.

“Thousands of developers around the globe already rely on Snowflake as their go-to development platform. Our latest innovations continue to push the boundaries of what builders can create in the AI Data Cloud, bringing familiar, yet powerful experiences to their enterprise data where it already lives,” said Jeff Hollan, head of applications and developer platform at Snowflake.

Offering these latest innovations, the company is “ultimately putting AI in the hands of more users,” he added.

It is also arming developers with more ways to accelerate their AI development directly on their data in the AI Data Cloud with Snowflake Notebooks (now in public preview) natively integrated with the full breadth of the Snowflake platform, including Snowpark ML, Streamlit and Snowflake Cortex AI. 

For example, Lukas Biewald, co-founder and CEO of Weights & Biases, shared that using Snowflake Notebooks, the company can easily integrate its experiment tracking with Weights & Biases directly within notebooks.

“Centralized, secure access to Snowflake data and compute lets you run data engineering and analysis alongside ML modeling in a notebook-style format that’s familiar, intuitive and powerful. We’re excited to see how our customers use it to run experiments faster,” Biewald said.

The company also revealed innovations and enhancements to Snowflake Cortex AI to unlock the next wave of enterprise AI for customers with efficient and trusted ways to create AI-powered applications.

The additions include new chat experiences to empower organizations to develop chatbots within minutes so they can talk directly to their enterprise data and get the answers they need faster.

Snowflake is also democratizing how users can customize AI for specific industry use cases through a new no-code interactive interface, access to industry-leading large language models and serverless fine-tunings.