Snowflake said, "Our product line, especially in the AI field, has been operating at full capacity. The era of enterprise AI has arrived, right here at Snowflake."

Snowflake's latest quarterly performance has pleased Wall Street. Snowflake has released sales forecasts for the current quarter that exceed expectations, indicating that its new AI-focused products are driving faster growth.

Snowflake announced that its product revenue for the current quarter (ending in July) is expected to be between $805 million and $810 million, a year-on-year growth of 34%. This exceeds analysts' forecasted revenue of $787.5 million. In addition, the company has raised its annual product sales forecast from $3.25 billion to $3.3 billion.

All about generative AI

Snowflake had negotiated to acquire generative AI startup Reka AI for over $1 billion, but the negotiations ended without an agreement. In April of this year, Snowflake launched its own large-scale language model suite, Arctic, which now allows customers to use third-party AI models such as Mistral and Meta on the company's platform called Snowflake Cortex to process their data.

In addition to the failed Reka AI deal, Snowflake also announced a definitive agreement to acquire TruEra, an AI startup focused on testing, debugging, and monitoring machine learning models and large-scale language model applications in production.

Snowflake announced plans to "acquire certain technology assets and hire key employees" from this AI startup, which raised $25 million in funding in 2022.

Regarding this acquisition, Anupam Datta, co-founder, president, and chief scientist of TruEra, said, "We look forward to entering a new phase with the Snowflake team, as we are both committed to delivering effective and trustworthy generative AI and predictive machine learning at a larger scale."

This acquisition marks Snowflake's sixth major investment in enhancing its data cloud capabilities and its third major move in the field of data observability. Previously, Snowflake had invested in two monitoring solution companies, Observe and Metaplane.

In a blog post, Snowflake emphasized that the acquisition of TruEra will help the company ensure more accurate and trustworthy data for training AI models.

TruEra has played an important role in addressing the black box problem of AI, and its team behind it is a solution expert based on RAG. After the acquisition, the three co-founders of TruEra will join Snowflake and work together with the TruEra team.

This AI-focused approach allows Snowflake to better compete with other companies in the field, such as Databricks, which offers similar services. Interestingly, since Databricks acquired MosaicML, it has also adopted a similar acquisition strategy.

Baris Gultekin, head of Snowflake's AI division, said that he has worked with Ramaswamy at Google for over 20 years and described him as an outstanding leader. "Sridhar has deep expertise in AI and data systems. He has managed large-scale data systems and AI systems at Google," Gultekin said.

Gultekin further stated that Snowflake is developing large-scale language models at a very affordable price and prioritizes the security of customer data. "Despite a 17-fold reduction in computing budget, Arctic is on par with Llama 3 70B in language understanding and reasoning, while surpassing enterprise benchmarks," Gultekin said.

In addition, Microsoft announced an expanded partnership with Snowflake to provide customers with a seamless data experience. As part of the collaboration, Microsoft Fabric's OneLake now supports Apache Iceberg and facilitates bi-directional data access between Snowflake and Fabric.

OneLake is a SaaS-based unified open data foundation launched by Microsoft with Fabric. This foundation highlights Microsoft's commitment to open standards. Microsoft Fabric OneLake's support for Iceberg, as well as its previous support for Delta Lake, further strengthens this commitment.

Essentially, Snowflake can store data in the format of Iceberg in OneLake. Data written by both Snowflake and Fabric can be accessed in Iceberg and Delta Lake formats through XTable transformations in OneLake. Snowflake can read any Fabric data asset in OneLake quickly, whether they are physically or virtually stored.

But that's not all.

In a recent interview, Ramaswamy revealed that the cloud data company plans to deepen its collaboration with AI giant NVIDIA. "We are collaborating with NVIDIA in multiple ways - our foundational model Arctic was unsurprisingly built on NVIDIA chips. There are many exciting things to look forward to, and of course, Jensen is a visionary in AI," Ramaswamy said.

Snowflake is expected to make more announcements at the Data Cloud Summit in June of this year. As Ramaswamy said, "Our product line, especially in the AI field, has been growing at a rapid pace. The era of enterprise AI has arrived, right here at Snowflake."