Main content

Snowflake CEO insists his company can take the heat

Jessica Twentyman Profile picture for user jtwentyman October 4, 2019
At the company’s Summit event in London this week, Snowflake’s Frank Slootman discussed what it takes to compete against, while relying on, the giants of public cloud

Frank Slootman

When companies are liberated from the constraints imposed by costly and hard-to-manage on-premise infrastructure, their explorations of data can really take off. That was the message from Frank Slootman, chairman and CEO of cloud-based data warehousing company Snowflake, speaking at the company’s Summit event in London.

Still privately held, Snowflake now has some 2,500 customers and is adding to the list at a rate of around 500 per quarter, he told attendees - not bad for a platform that was only launched to the public in 2015. But perhaps more importantly, once they’ve settled in on the Snowflake platform, those customers use it more and more. The average customer, according to Slootman, has grown its utilisation of Snowflake by 150% since first coming on board. Overall, Snowflake now handles 300 million queries per day from its customer base.

But public cloud infrastructure is vital to what Snowflake does - and to some extent, that creates friction between the company and its tech-giant hosting partners, who all offer data warehousing services of their own.

What’s clever about Snowflake is how it splits compute from storage. This means that it can spin up ‘virtual data warehouses’ to handle specific workloads, which can be scaled up and down according to need, enabling concurrent workloads to run without impacting each other.

That, of course, calls for massive infrastructure resources of the kind that only Amazon Web Services (AWS), Microsoft Azure and Google Cloud can provide. While Snowflake started life on AWS, which offers its own Redshift data warehousing service, it has subsequently launched on Azure (as of mid-2018), which offers Azure SQL Data Warehouse, and more recently, Google Cloud (as of June 2019), which offers BigQuery.

Competitive dynamics

In effect, that means that in many sales engagements, Snowflake is competing against one of these ‘own-brand’ cloud data warehousing services, or trying to persuade customers to migrate away from them, in favour of its own platform. Migration from Redshift to Snowflake, for example, has been a theme of several Snowflake customers stories covered in the past by diginomica, including at food delivery service Deliveroo, and at fitness-tracking applications Strava and ASICS-owned Runkeeper.

I was curious to ask Slootman how this situation of simultaneous dependence on, and competition with, the cloud giants is handled by his company - and wasn’t surprised to hear him concede that it can be tricky. He said:

It’s unavoidable for us to contest these first-party, first-generation cloud products. Does that create a tension? Yes, it does. But we are ourselves a very large customer of these cloud companies, first of all, because we pay them huge amounts of money. Secondly, we are a big partner, because we’re bringing very large, on-premise workloads to their cloud infrastructure, so we are net-benefiting them, right? And thirdly, yes, we do compete against them - and that creates a very complex dynamic and, depending on what day of the week it is, things can get a little testy. But for the most part, and I’ve been really involved in these situations over the last couple of months, we’re all adults, we all understand we have to live together and sometimes it’s very uncomfortable.

A great deal of effort goes into ironing out difficulties at quarterly business meetings with the cloud providers, he said, but the situation demands a clear-eyed, level-headed view of market trends and demand patterns:

This is the world we now live in. These are the new platforms. This is the new reality. Is it completely balanced and harmonious? No. But this is what it is, and by the way, the cloud vendors are committed to running very vibrant ecosystems. They want to, they have to. It’s no good to go to AWS and you only get one choice. That’s like going to Sainsburys and getting one brand of toothpaste. Customers won’t like that - and by the way, that particular toothpaste might not be very good quality.

In other words, in Slootman’s view, Amazon loves having Snowflake on board - but its for Redshift sales executives, not so much. Either way, the market for replacing first-generation cloud data warehouse deployments has proved fruitful for Snowflake, because these are customers already won over to the cloud concept. It’s currently a much bigger market for the company right now than replacing existing on-premise data warehouse infrastructures from the likes of Teradata, for example. In these conversations, the cloud issue continues to loom large for customers, Slootman says, although he’s optimistic that will change:

Going forwards, we’re going to be in tons and tons of on-premise conversations, where we’re lifting massive data estates to the cloud and it’s a huge undertaking but very, very transformational. But they all know they have to do it. Time is not their friend. They can’t stay where they are. All the new, interesting, compelling software is going to be on the public cloud. It’s not going to be on-premise.

A change of ownership?

All this doesn’t rule out an acquisition of Snowflake by one or other of the cloud giants - and at a hefty price, given the seven rounds of funding amounting to almost $929 million that the company has raised. In terms of a possible IPO, meanwhile, Slootman had this to say:

We know we don’t need to public to raise capital. The issue for an IPO is liquidity [...] but our investors are not pushing for the company to go public at all. They’d be happy to sit here for years and let us build the company. Where typically the pressure comes from is employees. We’ve been around for seven years, people have been here a long time and there’s always an implied social contract between the company and its employees that at some point their equity will be monetizable. That pressure does build up over time and we’re very mindful of that.

But given the compliance, administrative and regulatory effort involved in an IPO, he added, an offering wouldn’t be feasible for Snowflake any time before early summer of 2020, and the US presidential elections of that year look likely to create market uncertainty that would not be conducive, in any case.

For now, then, Slootman and his team are focusing their efforts on the company’s Snowflake Data Exchange, announced at the company’s annual user conference in June. This will enable employees at companies that use Snowflake to browse a private marketplace, a bit like an app store, to search for, access and query company data sets in an access-controlled way. The idea here is to turn them into self-servicing ‘data consumers’. At the same time, Snowflake is recruiting third-party data providers to offer their data on a public version of the exchange and so far around 150 have signed up, such as Weather Source. Said Slootman:

I’m of the opinion that the private data exchange model will eventually become the dominant way of deploying our technology. Data sharing is so important. Our whole theme here is about breaking down silos and erasing boundaries. The Data Exchange is the epitomy of that vision.

And that, he adds, can’t happen without the cloud - regardless of the industry in which a customer operates. The main mission for Snowflake, then, is hammering that message home. It’s a task that the straight-talking Slootman seems to relish:

When I talk to these big banks, they say ‘We’re just going to stay where we are.’ And I’m like, ‘You can’t stay where you are. You will move - it’s just a question of when, and how hard it’s going to be. Staying where you are is not an option. Get your head around that one.’


A grey colored placeholder image