Snowflake in ASEAN: The data layer beneath the AI boom

Southeast Asia's AI opportunity is partly constrained by the fragmented data companies are trying to run models on. The problem is becoming more important as firms move from experimenting with AI to putting it into the business. Snowflake is selling infrastructure for that problem. Its regional footprint now includes commercial AWS regions in Malaysia and Thailand, alongside Singapore and Jakarta.

Snowflake is not an AI company in the way some may mean the term. In its latest annual filing, Snowflake describes its product as solving "the decades-old problem of data silos and data governance." The software runs on Amazon Web Services, Microsoft Azure and Google Cloud. Snowflake does not own those clouds. It sells a layer on top of them: a place to bring data together, analyse it, share it and, increasingly, point AI at it.

Snowflake is not alone in selling this layer. It is competing with the companies that already sit underneath much of Southeast Asia's cloud infrastructure. AWS, Microsoft and Google can all offer customers their own data and AI services. Databricks takes a different approach to the same broad problem of bringing enterprise data and AI workloads together.

That leaves Snowflake with an unusual position. It runs across the major clouds rather than owning one of them, which gives it a different proposition for enterprises with mixed cloud environments. It also means Snowflake has to prove that its independent data layer is worth paying for when the hyperscalers increasingly offer overlapping services of their own.

What happened

When Snowflake opened a Jakarta public preview on 15 March 2023, the newsroom said the platform "helps organizations break down data silos." It also said that "with data residency legislation in Indonesia, the need to unify data across the ecosystem is now a business critical need."

Indonesia's 2019 electronic-systems regulation requires public-scope operators to keep systems and data in the country. Private operators may store or process data offshore if Indonesian supervision still works. A local region makes the proposition easier for buyers that need to keep systems or data in Indonesia. It does not mean every Indonesian company is subject to a blanket localisation requirement.

Snowflake's customer examples show what that layer looks like in practice.

Danny Thien, head of data at Singapore telco M1, said the company had "over 200 legacy databases" before it built a data lake on Snowflake, a central repository that brings data from different systems together for analysis and use. He described the problems as silos, duplication and records the business could not trust.

Another example is GXS Bank, the Grab-Singtel digital bank launched in August 2022 and licensed by the Monetary Authority of Singapore. The bank used Snowflake to bring together Oracle for finance, Salesforce for marketing and Axiom for regulatory reporting to MAS. In this case, Snowflake looks less like a dashboard and more like a control layer connecting systems that have different jobs.

Japfa, the Singapore-headquartered protein producer with operations in Indonesia, Vietnam, India, Myanmar and Bangladesh, represents a cross-border customer of Snowflake, facing data fragmented across systems spread across markets with different operating and regulatory environments. That makes the problem harder than simply consolidating databases in one country.

In the February 2026 earnings letter, Snowflake's chief executive, Sridhar Ramaswamy, described a decade spent building "the foundation that makes AI safe and scalable - a single source of truth, cross-cloud interoperability, and enterprise-grade governance." Agentic features, he said, sit "on top of that platform." The same release said more than 9,100 accounts across its global customer base were using Snowflake AI features.

What it means

A February 2026 survey of 330 executives, published by Singapore's Economic Development Board with McKinsey and Tech in Asia, listed fragmented data among the conditions slowing AI adoption. Talent, unclear returns and integration complexity ranked higher. About one in ten executives identified data quality and availability as a key barrier.

This is relevant for Snowflake because data is necessary for enterprise AI, but it is not the only bottleneck. A company can consolidate its databases and still lack the people to build useful applications, clean data to train or retrieve from, or a clear reason to deploy them.

Snowflake's opportunity is therefore not simply to help ASEAN companies prepare their data for AI. It is to become the layer through which that data continues to be used as companies move from analytics to AI.

Singapore's Infocomm Media Development Authority spent 2025 treating governed data as a precondition for trusted AI. Bank Negara Malaysia's 2019 outsourcing rules require banks using cloud services to consider sovereignty and geographically dispersed infrastructure. The constraints are different across ASEAN, but the direction is similar: where data sits, who can access it and how it is governed are becoming part of the AI architecture, not just IT administration.

That creates an opening for Snowflake. Its regional cloud footprint can help enterprises place data closer to where regulatory and operational requirements demand it, while its cross-cloud model gives it a role that does not depend on one hyperscaler.

But AI complicates the proposition. Buying the data layer is not the same as buying in-country inference. Snowflake Cortex, its suite of AI services for working with enterprise data, can use cross-region inference, depending on how the account is configured. Stored tables remain in the account region, but a prompt can be processed in another region without the underlying customer data being stored there.

For years, procurement teams could ask a relatively straightforward question about cloud data: where is it stored? As AI features sit on top of that data, the focus becomes more granular: where is it processed, which model sees it, and under what controls?

That is where Snowflake's independent position becomes both an opportunity and a risk. It can sit across AWS, Azure and Google Cloud, but those same companies increasingly offer their own data, analytics and AI layers. Snowflake has to make the layer in between valuable enough that customers keep paying for it.

What to watch

• Whether the independent data layer stays worth paying for. Snowflake's appeal is partly its ability to sit across clouds. If enterprises continue to pay for that independence, the model holds. If more customers consolidate their data and AI workloads with their existing cloud provider, Snowflake has a harder case to make.

• Whether ASEAN customers actually use the new local regions. Snowflake lists commercial AWS regions in Malaysia and Thailand, but the existence of a local region does not tell us who is using it. Named customers, particularly in regulated industries, would show whether local infrastructure is translating into actual demand.

• What happens to post-merger data estates. ASEAN's consolidation is creating companies with multiple systems, databases and cloud environments that now have to work as one. Indonesia's XLSmart is one example, following the combination of XL Axiata and Smartfren. Snowflake is one option, but post-merger rationalisation will show which platforms companies actually choose when they have to clean up years of accumulated technology.