South-east Asia has never produced an enterprise software giant. AI might change that.

FOR as long as venture capital has existed in South-east Asia, one piece of conventional wisdom has gone largely unquestioned: Do not build enterprise software in this region. The market is too small, as the argument goes.

And enterprises buy from global brands, so regional startups should stick to consumer apps and ride-hailing.

That view was never quite right. And today, it could be wrong.

The market was always there

The common objection is that the enterprise software market in this region is too small to sustain a major company. The spending data tells a different story.

Salesforce generated US$3.9 billion in revenue from the Asia-Pacific in its fiscal year 2025, a 12 per cent year-on-year increase that made it the company’s fastest-growing geography.

SAP’s Asia-Pacific and Japan division delivered 5.3 billion euros (S$7.9 billion) in total revenue, and ServiceNow achieved US$1.53 billion in 2025.

These are Asia-Pacific-wide figures that include Japan, Australia and India. But South-east Asia is the fastest-growing sub-region, and even its share represents a multi-billion-dollar opportunity.

The market was never too small. It was simply being served entirely by foreign incumbents.

The reason was straightforward: Local companies could not build products good enough to compete. Enterprise software demands deep systems architecture, rigorous security and compliance, and the kind of reliability CIOs are willing to stake their career on.

SEE ALSO

Creative, Razer, Secretlab – What’s the playbook for building a global brand from Singapore?

What building an AI tool taught me about venture capital

For two decades, the engineering talent to build at that level was concentrated in Silicon Valley, Seattle and Bengaluru.

Most of the South-east Asian startups that tried failed – not for lack of market demand, but because they could not close the product gap against incumbents with hundreds of engineers and decades of accumulated capability.

Three forces changing the equation

That product gap is now narrowing, driven by three forces.

First, AI has altered the economics of building software on two fronts. AI-powered coding tools mean a team of 10 engineers in Singapore can produce output that would have required 30 or 40, five years ago.

And the widespread availability of both frontier and open-source foundation models, from OpenAI and Anthropic to Meta’s Llama and Alibaba’s Qwen, means a startup no longer needs a 100-person machine learning team to embed world-class AI capabilities into its product. The building blocks are accessible to anyone.

To be clear, enterprise software is far more than code. It requires sales capability, security certifications, service-level agreement commitments, and the patience to survive long procurement cycles.

AI does not solve all of these. But engineering capacity was the binding constraint that kept South-east Asian teams from even reaching the starting line. That constraint has loosened significantly.

Second, the current wave of enterprise AI demands far more customisation than the software that preceded it. A customer relationship management or enterprise resource planning suite can be sold as a relatively standardised product across geographies.

An AI agent that processes insurance claims in Bahasa Indonesia, manages procurement for a Thai conglomerate with unique approval hierarchies, or automates compliance reporting under Singapore’s regulatory framework, cannot.

South-east Asia is not one market but a patchwork of distinct regulatory systems, business cultures and languages. That complexity, long seen as a liability, is precisely what makes it defensible against a one-size-fits-all product built in Palo Alto.

Third, founders building from Singapore have a structural go-to-market advantage that is easy to underestimate.

Singapore is the Asia-Pacific regional headquarters for Salesforce, SAP, Microsoft, Google and AWS, among others.

This has created an unusually deep concentration of enterprise sales talent: professionals who understand how to sell complex software to large organisations in the region, who have relationships with CIOs and procurement teams from Tokyo to Jakarta, and who know how enterprise buying decisions are made in each market.

Local founders who are well plugged into the ecosystem know who and where they can recruit from. And unlike in the US, where enterprise software is increasingly bought through self-serve channels, South-east Asia remains a market where relationships, channel partners and trusted advisors determine who gets a seat at the table.

The playbook: look outward, build locally, move first

For a South-east Asian enterprise company to succeed, however, it cannot simply build a cheaper, more localised version of Salesforce. “Cheaper and more localised” is not a durable advantage against an incumbent with US$38 billion in annual revenue and a brand every CIO already trusts.

What founders should do instead is look at the new categories of enterprise software emerging in the US and China, categories that AI has made possible for the first time, and ask: “How do I build this for my market, before they come here?

In the US, Harvey is reinventing legal workflows with AI agents. Legora just raised US$550 million for AI-native contract intelligence. Juno is automating tax preparation.

In China, companies are building agentic AI for supply chain orchestration and manufacturing quality control.

None of these are iterations on existing software. They are entirely new product categories, where it is increasingly targeting services markets, or what they call “outcome-as-a-service”.

Each one has a South-east Asian equivalent waiting to be built, adapted for local regulations, languages and business practices.

A sceptic might argue that AI also makes it easier for these US startups to localise into South-east Asia. That is true for standardised products.

But for the categories with the greatest defensibility, specifically those where AI agents codify the tacit knowledge that currently lives in the heads of an enterprise’s most experienced workers, remote localisation is not enough.

The procurement officer who knows which suppliers actually deliver on time, the compliance manager who understands how a regulation is enforced in practice, the operations lead who has memorised the unwritten approval chains – extracting and encoding their institutional knowledge requires deep, sustained, on-the-ground collaboration.

It cannot be done remotely or generically. And once codified, the switching cost is extraordinarily high. Replacing the system means re-extracting years of organisational intelligence from scratch.

This is a moat that a Western entrant cannot easily overcome, even with superior technology, because the value is not in the model. It is in the encoded context.

We are already seeing founders build in this direction. ReN3, a Singapore-based company, has built an agentic AI productivity platform purpose-built for accuracy, sovereign deployment and local regulations. It supports air-gapped on-premise environments and enables enterprises to create no-code AI agents for legal, finance, procurement, and HR, designed for the local regulations and data sovereignty requirements.

Level3AI builds AI agents that handle end-to-end customer engagement in voice, e-mail and chat for enterprises that include Carousell, GetGo and Carsome.

But what makes it distinctive is that it is not selling software. It is selling a measurable customer experience outcome, optimised for the Asia-Pacific’s linguistic and cultural complexity, from multilingual code-switching in Hong Kong to gender-appropriate forms of address in Malaysia.

Meanwhile, Calibrax.ai takes a complementary approach, building bespoke agentic workforces whose proprietary Business Context Protocol encodes an enterprise’s entities, workflows and decision logic. This is precisely the kind of deep embedding that creates durable switching costs.

The global incumbents are not standing still. And the US startups defining these new categories will eventually expand into the Asia-Pacific.

The founders who move now, embedding their agents into enterprise workflows before alternatives arrive, will have a compounding head start.

Those who wait may find the window has already closed. South-east Asia’s enterprise software market has always been large enough to sustain major companies. For the first time, the conditions exist for those companies to be built from within the region.

But the ones that succeed will need to work on the new category of software enabled by AI. And build it in this region first.

Chan Yip Pang is executive director of investment, and Isaac Kwa is associate in investment at Vertex Ventures South-east Asia and India