Snowflake Is Now A Fast Follower

If you missed Can Snowflake Capitalize In the Age of AI, here is the LINK. You can also find my old piece on SNOW from its IPO HERE.

Snowflake reported a good quarter. The key metrics of revenue growth, remaining performance obligations (RPO), and current RPO are accelerating. Free cash flow was better than expected.

In 1Q, Snowflake reported product revenues of $790 million. Reported revenues beat guidance and expectations by 5%. The revenue surprise (amount of beat) was the highest over the past two years.

This was the second quarter in a row where the Co. signed big deals. In this instance, Snowflake signed a $100 million deal.

RPO growth accelerated to 47% and CRPO growth increased to 31%,

Free cash flow was $331 million, a 40% FCF margin, and grew 16% on FCF/share.

Annual guidance was raised by $50 million, $10 million more than the 1Q beat, a positive sign for future revenue growth.

However, the Company reduced its operating margin guidance from 6% to 3% to account for higher GPU amortization costs. While the guidance is conservative, the market took this as a negative sign for future cash flow. The stock sold off.

Thoughts On The Business & CEO Change

Frank Slootman is a hall-of-fame Software CEO. The stock reacted negatively to Frank’s sudden retirement on the 4Q release.

However, innovation hasd slowed at Snowflake over the past 18 months. New products and features were taking longer to execute. Consensus was growing that in an AI-first world, Databricks was outexecuting Snowflake.

A product-led CEO of the caliber of Sridhar Ramaswamy is the right transition for SNOW. Sridhar led Google’s Advertising for multiple years, through challenging times such as the Mobile transition. Despite his successful 15-year stint at Google, he left Google to found Neeva. He intended to disrupt Search. He was brought on to Snowflake via Neeva’s acquihire. With Sridhar as CEO, the product velocity has already accelerated. Databricks is the leader. Snowflake is now a fast follower.

What is Snowflake Arctic?

Arctic is an open-sourced Medium Language Model focused on solving problems for Enterprises. Snow developed Arctic to help enterprises automate coding, SQL generation, and instruction following.

Arctic stacks up favorably to competitor models on the metric of enterprise intelligence for training and inference. It is built to do specific enterprise tasks, not for worldly wisdom. In layman terms, the model can code, generate SQL script, and follow instructions.

Snowflake has also provided the recipe documents publicly. Arctic is available on Snowflake Cortex, Hugging Face, AWS, Replicate, etc.

Snowflake built Arctic in three months, with a budget of $2 million. Competitor models are significantly more expensive to train.

You can read more about Arctic here.

How does building a Language Model help Snowflake?

The goal is to give customers recipes to train and deploy models cheaply. Enterprise customers evaluate projects based on Return on Investment (ROI). Therefore, the cost incurred is an important metric. Custom models, deployed on Snowflake, will enable Customers to execute faster. They can cleanse data, build migrations, create data pipelines, write SQL to transform data and build intelligent predictive applications faster. The low cost creates higher Customer ROI. Therefore, customers are likely to use the platform more. This is a win-win for both Snowflake and its customers.

Other Product Innovation

Steamlit is generally available this quarter. Streamlit enables clients to deploy Applications in Snowflake without other prep work.

Cortex AI is generally available this quarter. Cortex is the AI layer where customers can use language models to summarize and transform data, write programming code, write SQL queries using English prompts, and build Chatbots and Co-pilots. Most important, is that customers can deploy AI with a minimal budget.

Iceberg will be available later this year. Iceberg is a way that customers can use Snowflake to access and drive intelligence from data stored in Cloud storage (ex in AWS but not in Snowflake). For 2024, the Company’s guidance implies a headwind to revenues as some use cases imply some customers moving data out of Snowflake to reduce costs.

However, in the long run, Snowflake will benefit immensely from Iceberg as it opens all of the customer data, which is approximately 10-100X of what is stored within Snowflake.

Iceberg is also a competitive response to Databricks, which has a Data Lake architecture and is cheaper than Snowflake for Data Scientist purposes and applications.

Unistore, or hybrid tables, a Snowflake OLTP feature, will also be available later this year. I will expand on this later.

Today, the stock is implying a 21-22% 5-year revenue CAGR and a 35% FCF margin in 2030. Both assumptions, to me, are conservative. Remember, these are growth investments, and will not fit in a value framework. Snowflake will hold its annual product meeting in two weeks. I will go deeper into valuation, future growth, and new products at that time.

The Data cloud end market is $50-60 billion and growing rapidly. To avoid the capital costs of buying expensive GPUs and upgrading on-premises data centers, most enterprises need to move more data to Data Clouds such as Snowflake or Databricks. Together, these companies are 10% of this end market in 2024. There is plenty of growth for both in the future, as data is central to an AI strategy.

Thank you for reading!