How Snowflake is charting its growth in APAC

Named after a type of data architecture, Snowflake is synonymous with cloud-based data warehouses that organisations rely on to pull together data from multiple sources.

The company went public earlier this year, generating market interest in its technology that helps customers large and small glean business insights at a time when data has been billed by some as the new oil for the digital economy.

But Snowflake did not make its foray into Asia-Pacific (APAC) until about three years ago, when it opened an office in Sydney to tap the opportunities of Australia’s mature cloud market. This was followed by its subsequent expansion into Southeast Asia via Singapore.

In an interview with Computer Weekly, Geoff Soon, Snowflake’s managing director for South Asia, outlines the company’s growth strategy in the region, its relationship with hyper-scale cloud providers and how it cracked the enterprise segment which had been resistant to cloud-based data platforms.

How is Snowflake doing in Asia-Pacific? The company has seen some spectacular growth since it was founded in 2012, but what are some of your key priorities in this region?

Geoff Soon: We’re a relatively young company and we spent our first three years building a product that was launched only in 2015. Naturally, as a Silicon Valley-based company, our initial focus was on the US market where we found a strong niche in gaming that had a unique problem – to analyse certain structured and unstructured data. Our first 50 to 100 customers were predominantly in that space.

We came out to Asia-Pacific (APAC) around about three years ago, and we decided to base our APAC headquarters in Sydney. The reason we made that decision was because Australia was fairly advanced in its adoption of public cloud. As Snowflake is a public cloud company, it made a lot of sense to initially deploy in Australia.

However, we quickly recognised that there is an incredibly rich ecosystem in the ASEAN region. So, we set up our Singapore office around around 18 months ago and it has grown quite significantly. We have about a dozen people in Singapore, and more recently, I’ve been looking after the South Asian region, which includes India, ASEAN, Hong Kong, Taiwan and Macau.

Earlier in May, we deployed another team in India, which was challenging because it was in the middle of Covid-19. So, in terms of our priorities, we’re very much focused on the Singapore market, followed by India. We’re also seeing a lot of interest and success in Thailand, the Philippines and Malaysia.

Talk to me about this demand – what kinds of industries are using your solutions? What is the typical journey like for your customers? Have they been using Snowflake even before you set foot in the region?

Soon: It’s very much about building new relationships, but we have three unique go-to-market (GTM) or sales cycles. The first GTM is what we call digital natives or cloud-enabled organisations.

These are organisations that have almost no legacy infrastructure and were born in the cloud. They have typically attracted a couple of rounds of venture capital (VC) funding and, in some cases, they could be very large organisations. And because they have such a deep technical bench, once they make a decision, they come to us to do some final validation before they proceed to become a customer.

When people talk about data platforms, they get very wedded to what they currently do on-premise

So, we’ve had good success and publicly we can talk about some of our larger cloud-native customers such as Swiggy, India’s largest food delivery platform, which has been using Snowflake for about a year. In Singapore, we’ve also had a lot of fintech (financial technology) companies come onboard recently because we’ve got a lot of security certifications.

One thing about digital natives is that they don’t have existing legacy budgets, so they’re trying to find budget from their VC funding to spend on the platform. Because we offer a consumption model, you pay for what you use, making it very easy to go with Snowflake because you’re not paying expensive subscription fees or licence fees when you’re building your product. That’s why I think we’ve had a lot of success with digital natives.

We’ve also had a lot of success with small and medium-sized enterprises (SMEs). When you look at all the data platforms in the market, be it Oracle, Teradata or IBM, unless you have a couple of hundred thousand to about a million-plus dollars, there’s no way that you can afford their technology.

With Snowflake, you sign up on our website and we’ll give you a $400 credit. You can get going for thousands of dollars instead of hundreds of thousands or millions of dollars. We’ve been quite surprised by the number of SMEs that have signed up, but they aren’t looking necessarily for a data warehouse, they’re just looking for a platform to do analytics.

The final thing we’re doing that’s also the hardest is to break into the enterprise segment. I’ve been thrilled that after 12 to 15 months of hard work, we’re finally gaining good traction in the largest enterprises in Singapore, India and around Southeast Asia. The reason it’s so hard is because there are still a lot of workloads on-premise when it comes to data. So, first we had to convince those customers that they need to go to the cloud. And after we’ve convinced them to go to the cloud, we had to convince them to move their data to the cloud.

When people talk about data platforms, they get very wedded to what they currently do on-premise. But when you ask them what they are using for email, video conferencing or CRM (customer relationship management), they’ll say Gmail, Zoom or Salesforce. So, they are already in the cloud; it’s just that they don’t associate those workloads with traditional datacentre workloads.

We’ve been educating customers that they shouldn’t differentiate between different types of cloud workloads. They’re all data workloads and Snowflake, like Salesforce, ServiceNow or Zoom, is another highly secure software as a service (SaaS) that you can take advantage of.

There’s no doubt that they are data workloads, but there are also different classes of data workloads with different security requirements. Would you say that enterprises are still more conservative when it comes to putting mission-critical, sensitive data workloads in the cloud?

Soon: I agree with you that there’s different sensitivity of data. And that’s why we’ve invested so much time and effort in engineering to ensure that we meet all the relevant requirements. Snowflake is GDPR (General Data Protection Regulation) compliant, which means we’re able to achieve certification against Singapore’s PDPA (Personal Data Protection Act), which is fairly similar to GDPR.

When it comes to financial services and healthcare, Snowflake is PCI-DSS (Payment Card Industry Data Security Standard) and HIPAA (Health Insurance Portability and Accountability Act) compliant, respectively. It’s a lot of effort to get those certifications and get audited every year, but this will make it easier for healthcare providers to leverage a platform like Snowflake where we’ve done all the heavy lifting.

Snowflake depends on cloud vendors to expand its presence in this region. In Southeast Asia, where the major cloud suppliers have established cloud regions mostly in Singapore, has your expansion been limited in countries that are concerned about data sovereignty?

Soon: We made the choice to deploy our team in Singapore, knowing that the vast majority of public cloud workloads in Thailand, Philippines and Malaysia are run out of the Singapore datacentres of AWS (Amazon Web Services) and Microsoft Azure. That has helped to reduce our operating expenditure as we can use Singapore as a hub to reach other ASEAN markets. It would have been far more costly to deploy in four or five different datacentres.

Where we’re seeing a little bit of a challenge is Indonesia, which is starting to firm up its data sovereignty laws. And as you know, all three major cloud providers are making a concerted effort to establish their presence in Indonesia. Google Cloud has already opened a cloud region in Indonesia. I believe the other two are following suit very soon.

I understand that countries are looking to create data sovereignty requirements. But by the same token, many countries have also benefited massively by not doing so. For example, the emergence of the region’s two super apps, Gojek and Grab, has been helped by not having those requirements in every single country in ASEAN.

What is your relationship with the public cloud providers, which also offer their own cloud data warehouse services?

Soon: It’s an interesting situation because while they have specific products that compete with us, we’re also one of their largest customers. We’ve made substantial investments with each of these providers. With AWS specifically, it’s public that we’ve committed to spending $1.2bn with them over the next five years.

The interesting thing is that public cloud suppliers are incredibly diverse and complex in that they offer hundreds of services. We tend to get along extremely well with 90% of them. Let me give you some specific examples – the three major cloud providers are all trying to promote their marketplaces that bring together best-of-breed offerings to the market, and we do very substantial business through those marketplaces.

Where Snowflake really sees its role moving forward is being the Switzerland of data

The relationship also extends the benefits to customers when they procure Snowflake via these marketplaces. When I look at the cloud hyper-scalers, it’s really a battle for wallet share. If they can provide Snowflake on their platform, they are achieving that ultimate outcome.

While there will always be instances of friction, we see more opportunities to collaborate. One thing that Microsoft and Amazon have been very public about is their obsession about the customer. So, if the customer has a clear preference for Snowflake, the relationship tends to be smooth sailing.

Snowflake recently added some new capabilities, particularly integrations with Salesforce as well as some security and marketplace features. What is the thinking around the company’s product roadmap? Will there be more features that encompass a broader range of activities across the data lifecycle?

Soon: That’s a fantastic question. The recent integrations that we’ve announced with Salesforce is all about our overall strategy and vision of developing a data cloud, and I want to explain what we mean by that.

At the moment, our large enterprise customers’ data is still siloed. But it’s a little different than it was five to 10 years ago, where they had data stuck in their mainframe and client-server applications. If you look at where data is siloed today, it’s between application clouds like Salesforce, ServiceNow and SuccessFactors.

Then there are also organisations that are investing massively to build artificial intelligence and business applications on public cloud infrastructure, and they still have traditional on-premise systems as well.

Where Snowflake really sees its role moving forward is being the Switzerland of data where we seamlessly allow you to aggregate your data from all different SaaS offerings, your applications sitting on public cloud infrastructure and your traditional on-premise infrastructure.

But the most exciting thing is that while the data within your organisation is valuable, what we’ve seen with Covid-19 and its disruption to supply chains is that we need external data. That’s why Snowflake has just announced it has signed up its 100th major data provider. With external data, whether it’s from S&P Global or Crunchbase, you can start to create insights.

The second thing is that while Snowflake is best known for being a cloud data warehouse, we now have six very distinct workloads, all working off the same unified architecture and platform.

Our second largest workload after data warehousing is data engineering, which sounds very sexy but it’s not. It’s taking one type of data, and manipulating, massaging and transforming it to make it more accessible to analysts or data scientists.

Traditionally, people were taking very complex approaches to data engineering using open-source tools such as Hadoop and Spark. That requires you to learn languages like Python and R to manipulate the data and could take thousands of lines of code to get a transformation done.

Snowflake, however, predominantly uses SQL, one of the most accessible languages. When it comes to data and analytics, people are using SQL commands to do joins and manipulations to get the data into an accessible format.

Another workload that’s really started to skyrocket is the use of Snowflake not only as a data warehouse, but also a data lake. That’s because Snowflake can natively ingest and immediately analyse semi-structured data like IoT (internet of things) records and other machine-generated data. More people are now using Snowflake as a central repository for machine learning and data science initiatives.

The final two workloads are from SaaS providers that are building applications on top of Snowflake and using us as the data platform, as well as our data exchange.