Snowflake continues investing in growth in an AI revolution

[SINGAPORE] Artificial intelligence (AI) has a symbiotic relationship with data – and data management platform Snowflake has taken this ball and run with it.

From leveraging cloud computing for data analysis to now managing both structured and unstructured data for companies, Snowflake has evolved its business. Structured data is information that fits neatly into formats that machines can understand; unstructured data is information that does not.

Although revenue jumped 29 per cent to US$3.6 billion for FY2025 ended Jan 31 from US$2.8 billion the year before, the company’s losses are widening. Snowflake reported a loss of US$1.3 billion in FY2025, surpassing the US$836.1 million loss in FY2024.

But co-founder Benoit Dageville seems unfazed. He told The Business Times that this is the price of being in a high-growth company, with capital continually ploughed back into investing for the future.

“The reason we are growing is that we are investing; the day we stop investing in the future, our growth will stop very quickly and that’s bad,” he said.

Rather than focusing on profits, Snowflake is working on its free cash flow – the cash a company generates after cash outflows – with an eye trained on growing the business. Free cash flow for FY2025 was US$884.1 million, up from US$778.9 million in FY2024.

Snowflake also has its eye on expanding its investments in markets like South-east Asia. From the initial entry into Singapore in 2019, the company has expanded into Indonesia, Thailand, Malaysia and the Philippines.

But there are some regulatory issues on the horizon for data. Countries are enacting data sovereignty laws, under which customer data cannot leave the physical geography of the country. In Indonesia, for example, the data sovereignty law, the Personal Data Protection Law, took effect in October 2022.

Snowflake lets its customers decide where to locate their data, and how the data moves about, said co-founder Thierry Cruanes.

“This is critical because globalisation is freezing today, and you see concerns about sovereignty of data in many countries rising,” he said.

The biggest challenge right now lies in growing and scaling the data platform to meet customers’ needs at a price point that is not exorbitant. More automation is on the cards, said Dageville, which would keep these costs down.

Then there is also the growing data universe, as AI is able to ingest and crunch more types of data, requiring Snowflake to add ever more data. This requires a balancing act between adding more data products or dimensions without introducing too much complexity, said Cruanes.

Recent moves by the US government to limit computational power outside of Tier One countries do not trouble Snowflake for now. (There are about 18 first-world countries, including Japan, Britain, South Korea and the Netherlands, in Tier One.)

But DeepSeek has shown the way to build AI models, even without access to enormous amounts of computational power – and this will trigger innovation towards more efficiency.

Dageville said: “I would say it’s almost an opportunity for us, because (of) the way we leverage resources, and we are really optimising (our operations).”

Snowflake is undertaking research into how to efficiently leverage graphic processing units. When there are limits, people will find ways to innovate through it, he said.

The intersection of data and AI will continue to be a bright spot for Snowflake, as the company seeks to continue integrating technology for customers to leverage.

“The last thing you want is to take your data and send it to a (large language model) which is running god-knows-where. You want to always bring the code closer to the data and share the expertise in a secure way,” said Dageville.