AI makes processing unstructured data possible: Snowflake Co-founders

Building a good product is more important than being a chief executive officer (CEO), the two men said

Snowflake is a data AI cloud computing firm based in Montana, United States. It began as a startup and is now a listed company with a market capitalisation of $53 billion. BENOIT DAGEVILLE, co-founder and president of products of Snowflake, and THIERRY CRUANES, another co-founder, visited India for the first time recently. Building a good product is more important than being a chief executive officer (CEO), the two men told Shivani Shinde in an interview in Mumbai. Edited excerpts:

 

How significant is India for Snowflake?

 

Benoit: India is critical for two reasons: customer base and developer ecosystem. India is one of the fastest-growing regions for Snowflake, and we now have 600 employees here. It is also one of the regions with the most developers; a lot of software is built here. We’ve [Snowflake] been in India for five-and-a-half years, initially focusing on digital-native companies. Now, we’re expanding to enterprises in BFSI [banking, financial services, and insurance], manufacturing, and retail while also working with SMBs [small and medium businesses].

 

There is talk of chief information officers considering on-premises tech infrastructure. How do you view this trend?

 

Benoit: I am a firm believer in Cloud technology, and I don't think going back to on-premises solutions is a good idea. While concerns around security and control exist, these can be effectively addressed through Cloud solutions that offer sovereignty and strong security measures.

 

The notion that organisations can achieve better security and efficiency by managing their own data centres is somewhat misguided. The power of the Cloud lies in its ability to democratise technology, making advanced infrastructure accessible to everyone. If we were to return to on-premise models, only companies with vast financial and technical resources would be able to manage large-scale, elastic data centres. That would limit innovation to a select few.

 

Some say AI is shifting software to services-as-a-software business. What is your view?

 

Benoit: Some people assume that AI will completely disrupt software as we know it, largely because of models like ChatGPT and their impact on traditional workflows. But in the context of data, it’s important to recognise that AI is fundamentally a processing engine — especially for unstructured data.

 

When we started Snowflake, our goal was to build a platform for structured data and semi-structured data. Unstructured data has been a challenge — not because it’s difficult to store (storing files is easy), but because processing and extracting insights from unstructured data traditionally required human intervention. For the first time, AI changes that. It allows us to process unstructured data in ways that were never possible before. Now, businesses can extract insights from audio files, customer service transcripts, and other complex data formats. AI can convert voice recordings into text, analyse patterns, and generate dashboards in real time. This ability to process and unlock value from unstructured data is a game-changer for the data ecosystem.

 

Coming to the discussion about AI leading to service-as-a-software: I don’t see AI replacing core data platforms or the way applications are built. The underlying data foundation remains critical — AI sits on top of it, making insights more accessible rather than replacing traditional data structures.

 

Snowflake is one of the biggest successes in Cloud computing. What gave you the confidence to compete against industry giants?

 

Benoit: If we go back to when we started Snowflake in 2012, there were two major revolutions happening. One was the rise of Cloud computing and how to fully leverage it. At the time, many believed that storing large amounts of data on the Cloud wasn’t practical, particularly for enterprise applications. But we saw the potential — Cloud computing offered elasticity, scalability, and efficiency that traditional systems couldn’t match.

 

Our vision was to create a platform where businesses could consolidate all their data in one place, eliminating silos. But making that a reality required solving a key challenge. If all your data is in one place, your compute power needs to be there too. Only the Cloud could provide the elasticity required for this, allowing workloads to dynamically scale and compute resources to be allocated on demand.

 

Another major opportunity we saw was in offering data solutions as-a-service. We focused on creating a platform that was simple to use but extremely powerful, allowing any company regardless of size to access enterprise-grade data capabilities.

 

The other big shift at the time, which is still relevant today, was the explosion of semi-structured data that was machine-generated. Hadoop had emerged as a solution, but it wasn’t ideal — it was too complicated for most organisations to adopt effectively. We saw an opportunity to build something significantly better: A true Cloud-native data warehouse that could handle vast amounts of data efficiently while delivering real-time insights.

 

With Snowflake, we wanted to offer businesses the ability to query massive datasets instantly — moving from waiting for hours or even days for reports to getting answers in seconds.

 

When you talk about a full-fledged Cloud platform, have you reached the goal you set for yourself?

 

Thierry: It will always be a work in progress because the needs of a Cloud platform are constantly evolving. Technology evolves, and we continuously innovate to meet new demands. Our goal is to empower developers by offering an integrated platform that allows them to build applications quickly and efficiently.

 

Benoit: Integrating these components directly into the platform eliminates the need for developers to build everything from scratch. For example, if an application requires monitoring, these capabilities should be built into the platform rather than requiring developers to integrate them separately. AI can also be embedded within applications, but when provided at the platform level, it simplifies the development process significantly.

 

Ultimately, our goal is to make it easier for developers to build applications quickly and efficiently, leveraging the full potential of the Cloud.

 

You’ve chosen to retain the roles of chief technology officer (CTO) and head of product rather than stepping into the CEO role. Was that a conscious decision?

 

Benoit: Yes, it was a very conscious decision.

 

I often say that when people ask us about running a business, we respond: "No, we never set out to build a company”. Snowflake, as a company, was almost a collateral outcome of our true goal — we wanted to build an amazing product. That has always been our core expertise, and it's crucial to recognize where your strengths lie.

 

Our unique expertise is in building software, and I believe we’re very good at that. However, we’re not as skilled at managing a company—and more importantly, we have no desire to do so.

 

From day one, we decided that neither of us would take on the CEO role. Our venture capitalist, Mike Speiser, served as CEO initially, followed by Bob Muglia and then Frank Slootman. We recognized that the CEO isn’t the only crucial role in a company: What truly matters is assembling a leadership team that complements each other’s strengths. Empowering people with expertise that we don’t have is what ultimately enables a company to scale.

 

This approach is also deeply embedded in Snowflake’s culture. We understand that success comes from leveraging diverse skill sets rather than concentrating all decisions within one person.

 

It is said that AI is prompting founders to give away the CEO role and become CTOs or product heads. What’s your comment?

 

Thierry: I don’t know if AI is directly responsible for this trend, but I do believe that product development is more critical than ever. Snowflake was built around its product, and everything else — its success, its growth — has been a result of that.

 

A great CEO for a company like ours is someone who is deeply product- and technology-focused because everything stems from that foundation. It’s not about choosing between being a business leader and a product visionary — both are essential. That’s why, even today, I continue to write code.