East Asia’s crypto market is no longer moving as one bloc. New regional data shows a fragmented landscape where South Korea is trading artificial intelligence-linked tokens at scale, Hong Kong is pulling in institutional capital, Japan’s retail users are quietly moving on-chain, and China’s stablecoin activity is expanding despite official restrictions.

The region’s overall crypto economy contracted modestly between July 2025 and June 2026, in line with the global bear market. But the headline decline masks sharp differences between markets. South Korea remained the region’s largest crypto economy at US$449.1 billion, followed by Japan at US$228.3 billion, Hong Kong at US$192.2 billion, China at US$176.3 billion, and Taiwan at US$140.4 billion.

Also Read: Taiwan’s stablecoin moment: Why the NTD could outshine the dollar

For Southeast Asian founders, exchanges and regulators, the message is clear: crypto adoption is increasingly shaped by local market structure rather than broad regional sentiment. Tax policy, licensing, capital controls, retail culture and institutional access are now producing very different outcomes across Asia.

As Daniel Kim, CEO of Tiger Research, put it, “Asia is that rare region with both grassroots retail depth and institutional firepower.”

South Korea turns the AI trade into a crypto trade

South Korea’s crypto economy grew 12.3 per cent during the period, helped by an additional US$51.1 billion in exchange-related flows. Unlike Hong Kong, where institutions drive much of the activity, Korea’s market remains heavily retail-led.

That matters because Korean retail investors have a long record of chasing high-conviction, high-volatility themes. In equities, the country’s AI trade has centred partly on SK Hynix, whose memory chips are critical to data centres. In crypto, the same appetite has spilled into tokens linked to AI projects and infrastructure.

By June 2026, AI-related cryptocurrencies were the most popular thematic category in won-denominated trading volume, surpassing payment tokens such as XRP. This made South Korea a global outlier. AI-crypto trading reached only 0.91 per cent of yen-denominated activity in June 2026, while the equivalent share in won was 19.5 times higher.

The names changed quickly. Worldcoin led with US$7.41 billion in volume, followed by SAHARA at US$3.2 billion, VIRTUAL at US$2.7 billion, BIO at US$2 billion and NEAR at US$1.7 billion.

The turnover suggests Korean traders are not simply buying a sector and holding it. They are rotating rapidly through whichever token best captures the current AI narrative.

Also Read: Japan shows how non-USD stablecoins complement USDC and USDT

Korea’s regulatory environment has helped sustain that retail dominance. Crypto profits remained untaxed during the study period, although a delayed 22 per cent tax is scheduled to begin in 2027. At the same time, corporate participation is only gradually opening up after long-standing restrictions.

Francis Kang, Executive Director of Korea Blockchain Week, described institutions as being in “preparatory mode”, with major banks and securities firms running pilots around stablecoins, tokenisation and custody. If the tax is implemented while corporate access expands, Korea’s market could begin shifting from retail momentum to a more balanced structure.

Japan’s retail market is more active than it looks

Japan is often viewed as an institution-first crypto market because of its strict licensing regime and the growing role of banks and financial firms. Yet the data points to a more active consumer base than that reputation suggests.

The country recorded a US$228.3 billion crypto economy during the period. Decentralised exchanges, or DEXs, accounted for nearly 35 per cent of Japan’s service activity, the highest share among mature centralised-exchange markets in East Asia. DEX activity has risen more than 200 per cent since 2022, while centralised-exchange activity has been broadly flat.

Much of this is happening in smaller, retail-sized transactions. About 65.7 per cent of Japanese retail DEX swaps were in the US$10 to US$1,000 range. Roughly one in four users who withdrew funds from exchanges operating in Japan later deposited into decentralised finance protocols.

Taishi Sato, CEO of DeFimans, a subsidiary of SBI, said the most visible activity is in perpetual futures, a type of derivative contract with no expiry date. He linked this to Japan’s large base of foreign-exchange traders, who are already used to hedging macro exposure.

Still, tax has been a major brake. Japanese crypto traders faced a maximum marginal rate of 55 per cent during the study period. Reforms advanced in July 2026 could shift eligible gains towards separate taxation of about 20 per cent. If implemented, that could bring more retail volume onshore and into regulated venues.

For Southeast Asia, where retail crypto participation is high but tax treatment remains uneven, Japan’s case is worth watching. A clearer, lighter regime could show whether tax reform moves users back into compliant platforms rather than offshore or peer-to-peer channels.

Hong Kong builds the institutional layer

Hong Kong’s crypto economy, at US$192.2 billion, is defined less by retail speculation and more by its role as an institutional settlement hub.

Also Read: Nearly half of Asia Pacific consumers open to stablecoins within five years, Visa says

Institutional platforms, including over-the-counter desks, custodians and market makers, captured 16 per cent of Hong Kong’s service inflows during the period, up from around nine per cent two years earlier. No other East Asian market exceeded six per cent.

The city drew nearly US$24 billion in inbound service-to-service transfers, around six times Japan’s figure and 44 times South Korea’s. Outbound flows reached about US$11 billion, suggesting Hong Kong is functioning as a two-way corridor for institutional capital.

This is the result of policy design. Hong Kong has made stablecoins, tokenised finance and central bank digital currencies part of its financial strategy. It issued its first stablecoin licences in 2026 and has moved into market testing with institutional players.

For Singapore and other Southeast Asian financial centres, Hong Kong’s trajectory is a direct benchmark. Both cities want to attract serious digital-asset businesses without reopening the excesses of the last crypto cycle. The lesson so far is that institutional money appears willing to move into markets where licensing is strict but workable.

China shows prohibition does not erase demand

China remains the hardest market to measure because crypto services are officially banned. Even so, the report estimates its crypto economy at no less than US$176.3 billion.

Unlike its neighbours, China’s activity is dominated by peer-to-peer flows rather than exchanges. Domestic peer-to-peer activity accounted for 59.1 per cent of the country’s all-in crypto economy, a 3.5-fold increase in share over the prior period.

Stablecoins appear to be central to this shift. The number of unique wallets sending peer-to-peer stablecoin transactions grew 43-fold between the first quarter of 2024 and the second quarter of 2026. China’s self-custodied stablecoin holdings also turned over 33.2 times a year, more than three times the global average of 9.3 times.

That high velocity suggests stablecoins are being used less as a store of value and more as working capital or payment rails. The report notes a possible link with the expansion of China’s social credit system into finance and the internet in March 2025, though the relationship remains a hypothesis rather than proven causation.

The broader takeaway is familiar across emerging markets, including parts of Southeast Asia: when users face friction in formal financial channels, crypto activity may not disappear. It may simply move into less visible peer-to-peer networks.

Also Read: SEA’s stablecoin boom has a dollarisation problem nobody’s pricing in

East Asia now offers four different crypto futures at once. Korea shows the power and risk of retail narratives. Japan shows how tax can suppress an otherwise sophisticated user base. Hong Kong shows institutional capital following regulatory clarity. China shows the limits of prohibition.

For Southeast Asia, the question is no longer whether crypto adoption will continue, but what kind of adoption its rules will encourage.

The post East Asia’s crypto market splits as Korea bets on AI tokens and Hong Kong courts institutions appeared first on e27.

Across Southeast Asia’s technology hubs, engineering leaders are getting used to a new kind of productivity curve. In Singapore, Jakarta, Manila and Ho Chi Minh City, software teams are shipping faster, clearing backlogs more quickly and leaning on AI tools for work that once consumed hours of developer time.

The latest shift is not just about coding copilots suggesting the next line of code. The region is moving into the age of autonomous AI agents: tools that can generate tests, review pull requests, draft documentation, refactor code and execute multi-step engineering tasks with limited human input.

Also Read: I stopped hiring. I’m not sure it’s a strategy yet

On the surface, this looks like an unambiguous win for startups. According to the Agoda AI Developer Report 2026, which surveyed more than 800 developers and engineering executives across Southeast Asia and India, 55 per cent of developers now save at least seven hours a week using AI, up from 18 per cent a year earlier.

For founders managing tight runways, that kind of efficiency is hard to ignore. But beneath the productivity gains is a more difficult question: if AI absorbs the routine work that once trained junior developers, how will the region produce its next generation of senior engineers?

The rise of the super-senior engineer

Software teams used to scale in a fairly predictable way. As product demands grew, companies added headcount. Junior developers handled simpler tickets, built UI components, wrote tests, fixed bugs and learned under the supervision of more experienced engineers. Mid-level engineers took on larger features. Senior engineers reviewed architecture, guided technical decisions and caught mistakes before they reached production.

AI agents are weakening that link between output and team size.

“One highly experienced engineer can now achieve the output that previously required a team of five to ten engineers,” said Tajrij Kawakibi, CTO at Indonesian software firm PT. Quadra Konten Persada, in the Agoda report.

Jerome Asuncion, Head of Data and Engineering at LiVeritas Philippines, made a similar point, saying agentic AI has enabled his lean team to tackle initiatives that would once have required far more engineering capacity.

That is changing how Southeast Asian startups think about team design. Only 14 per cent of developers surveyed expect software engineering team structures to remain unchanged over the next 12 months. Twenty-nine per cent expect fewer engineers to manage a growing matrix of AI workflows. Another 25 per cent expect AI agents to become autonomous team members, and 23 per cent expect smaller, hyper-productive squads.

Also Read: AI startups are hiring around answers they haven’t earned yet

For early-stage startups, the calculation can seem brutal. Why hire several junior developers who need onboarding, mentoring and code review when one senior engineer with the right AI workflow can produce similar output more quickly?

Junior anxiety, executive optimism

The tension shows up clearly in how different levels of the engineering workforce view their own future.

Across the survey, 37 per cent of developers believe AI agents will make their roles less secure, while 32 per cent expect no change and 31 per cent believe their roles will become more secure. But the anxiety is not evenly distributed.

Among junior developers, 49 per cent feel less secure about their future. That falls to 30 per cent among mid-level developers and just 17 per cent among CTOs and VPs of Engineering.

The gap reflects what each group sees in day-to-day work. Senior leaders experience AI as leverage. It removes repetitive implementation work and gives them more time for architecture, risk management and product trade-offs. The report found that 83 per cent of CTOs and VPs are already using production-grade AI agents, compared with only 33 per cent of junior developers.

For juniors, the picture is less reassuring. The very tasks they were once given to build confidence (writing basic documentation, generating unit tests, refactoring small functions, fixing minor bugs) are among the first to be automated. Junior developers are also nearly three times as likely as senior engineers to cite lack of skills as their main operational barrier.

This is not simply a labour market issue. It is a training issue.

Skills are moving up the stack

As AI takes on more code production, the definition of engineering value is changing. Syntax still matters, but it is no longer the main differentiator. The higher-value work now sits in system design, judgement, orchestration and review.

Developers surveyed by Agoda ranked system architecture and design as the most important future skill, cited by 66 per cent of respondents. AI literacy came next at 61 per cent, covering the ability to understand model behaviour, context limits, prompting, failure modes and when not to trust outputs. Agent orchestration and management followed at 51 per cent.

Traditional coding fundamentals, by contrast, were selected by only 18 per cent.

Also Read: Vietnam’s returning engineers are high-quality talent. Keeping them is the real problem

That does not mean coding basics are irrelevant. A developer who cannot read code, reason through execution paths or understand data flows will struggle to evaluate AI-generated work. But the premium is shifting from producing code manually to knowing what should be built, how systems might fail and whether an automated output is safe to deploy.

Developers appear to understand this. Forty-three per cent are actively building new technical skills, while 37 per cent are trying to shift their daily work towards higher-value tasks beyond raw code production.

For Southeast Asia, where many startups have historically relied on young engineering talent to scale affordably, this shift could be disruptive. The region does not just need more developers; it needs developers who can grow into architects, security-minded reviewers and technical leaders.

The mentorship gap

The hard part is that engineering judgement has traditionally been built through repetition. Senior developers did not become senior by skipping the mundane work. They wrote imperfect code, broke things, handled edge cases, received blunt pull request feedback and slowly developed intuition.

If junior developers use AI agents to generate most of their code from day one, they may move faster but learn less deeply. They may not develop the instinct to spot a subtle security vulnerability, a flawed database design, a hallucinated dependency or an architectural shortcut that will become expensive later.

M. Ridwan Agustiawan of dataxet described this as a “judgment gap” in the report: less experienced developers may either accept AI outputs too readily or fail to recognise when human validation is essential.

That risk is already visible in the data. Eighty-six per cent of developers agree that human review of AI outputs remains essential, and 42 per cent say the individual developer should carry primary responsibility for outages caused by AI agents.

In other words, AI can write the code, but humans still own the consequences.

What startups should do now

Southeast Asian startups cannot afford to stop hiring juniors. Doing so may improve short-term productivity but create a leadership shortage later. Instead, CTOs and engineering managers need to redesign how junior talent is selected and trained.

First, hiring should move beyond syntax-heavy tests. Candidates still need fundamentals, but interviews should also assess whether they can debug AI-generated code, explain trade-offs and question flawed assumptions.

Second, mentorship should become more deliberate. Junior developers should be paired with senior engineers during AI-assisted workflows, not only after code is written. The key lesson is no longer just “how do you write this function?” but “why is this the right design, what could go wrong, and how do we verify the agent’s output?”

Also Read: AI has answers, experience has judgment

Third, companies should create safe ownership zones. Juniors can manage low-risk AI workflows such as test generation, internal tooling or documentation automation, giving them room to make mistakes without threatening production systems.

AI tools will soon be widely available to every serious engineering team. The real advantage will belong to companies that know how to turn those tools into learning systems.

For Southeast Asia’s startup ecosystem, the junior developer dilemma is not a reason to retreat from AI. It is a warning that productivity without mentorship can hollow out the talent pipeline. The startups that solve this early will not just ship faster; they will build the next generation of engineering leaders.

The post The junior developer dilemma: How AI is reshaping tech talent in Southeast Asia appeared first on e27.

Selecting a scalable enterprise solution is no longer just an operational choice; it is a core business strategy. For small and medium businesses navigating a highly digitalized economy, legacy architectures fail to meet modern efficiency standards. Integrating advanced automation into financial workflows is crucial for maintaining agility and staying competitive in a rapidly evolving market.

Modern Tech Frameworks: MCP and Agentic AI

The integration of Model Context Protocol (MCP) and Agentic AI represents a massive leap forward for enterprise software. Static systems require rigid, manual configurations to pass data between an AI model and internal database structures. Dynamic MCP, however, maps entire corporate memory frameworks automatically. This allows autonomous agents to execute complex, contextual financial workflows with zero human intervention.

For small and medium enterprises, this technical shift drastically minimizes processing bottlenecks. Instead of human operators spending hours reconciling accounts or chasing cross-border invoice discrepancies, certified AI agents securely interact with real-time operational data. This ensures absolute workflow compliance and continuous strategic optimization.

 

The Steep Financial Cost of Rigid, Legacy Platforms

Sticking with low-cost, off-the-shelf software with no room for no-code customization or modern MCP adoption carries major hidden expenses. While the initial subscription price might look attractive, these rigid systems quickly create operational barriers.

• Exponential Integration Overheads:Connecting third-party productivity tools or proprietary operational modules requires custom software development, which adds significant recurring IT costs.
• Massive API Consumption Fees:Legacy architectures rely on heavy, unoptimized API calls. Without natively integrated infrastructure, routing data out to LLMs drains tokens and leads to massive cloud computing invoices.
• Severe Operational Stagnation: When business models evolve, a rigid platform traps the enterprise in slow, manual workflows. This completely wipes out the initial savings of choosing a cheaper option.

Also Read: Top 3 popular GEO monitoring tool for SEO optimisation targeting service industry in Singapore

Evaluating the Top 5 AI Accounting ERP for Modern SMBs

Navigating the enterprise application market requires balancing immediate software functionality against long-term operational costs and architectural flexibility. Below is an analytical review of the leading solutions tailored for small and medium operations.

Multiable

• Pros:

• • Features over 500 agent-ready APIs that cut agentic AI token costs by over 90% compared to systems without built-in API frameworks.
  • Incorporates an advanced dynamic MCP covering standard features as well as custom-made functions configured or developed directly by the user.
  • Operates on a highly optimized, lightweight Linux-based cloud infrastructure to minimize underlying hardware overheads.
  • Delivers comprehensive multi-currency tracking along with native, real-time financial consolidated reporting.
  • Users can access the platform at Multiable to seamlessly integrate their entire front-end and back-end operational workflows.
• Cons:
  • Support services requested during weekends or public holidays will incur extra charges.
  • The pricing model may be out of touch for small mom-and-pop businesses with fewer than 10 staff members.
  • Customizations made outside the standard no-code platform require specialized internal training.

Evaluation Summary:
By delivering a vast array of pre-built APIs along with an adaptive database architecture, Multiable ERP slashes the hidden costs of AI integration. It stands out as the single best AI Accounting ERP for SMB in Singapore, particularly for scaling organizations that require complete custom flexibility without paying massive premium fees.

Chillaccount

• Pros:

• • Built from the ground up as a cloud-native platform tailored for agile, rapid deployments.
  • Features a highly intuitive financial ledger configuration designed to simplify everyday corporate book-keeping.
  • The digital interface at Chillaccountoffers straightforward automated invoicing right out of the box.
• Cons:
  • Completely lacks deep supply chain or advanced manufacturing module integrations.
  • Offers limited multi-company consolidation capabilities for complex, scaling corporate groups.
  • The platform lacks a robust, built-in dynamic MCP architecture to handle complex external AI orchestration.

Evaluation Summary:
A clean, cloud-native design ensures that early-stage businesses can automate their core accounting books quickly. It serves as a dependable starting option for straightforward operations that do not yet face intricate supply chain or multi-tiered corporate tracking challenges.

Microsoft Dynamics 365

• Pros:

• • Native, seamless data connectivity across the entire Microsoft 365 product ecosystem.
  • Highly scalable framework capable of managing huge multi-national operations.
  • Offers extensive global compliance reports that cover a massive variety of regulatory jurisdictions.
• Cons:
  • The resource-hungry Windows Server operating system requirement means hardware and infrastructure costs can be up to 10x higher than Linux-based solutions.
  • Real-time performance issues and query latency within AzureSQL remain a noted concern during heavy data loads.
  • Implementation paths are exceptionally complex and require highly paid external consultants.
  • The subscription matrix features convoluted, multi-tiered licensing structures that escalate quickly.

Evaluation Summary:
Large businesses benefit significantly from deep ecosystem ties and extensive global compliance reporting. However, smaller growing firms must carefully balance these features against the high infrastructure costs and complex setup demands of this heavy environment.

Oracle NetSuite

• Pros:

• • Provides a highly comprehensive cloud-based single database architecture for unified corporate tracking.
  • Features powerful, deeply detailed business intelligence dashboards.
  • Strong global tax management engines built directly into the core platform.
• Cons:
  • Customers frequently face steep increments in SaaS fees upon renewal, which can hit up to 50% of the first SaaS contract price.
  • Suffers from a complete lack of built-in MES support, forcing firms to rely on clumsy third-party integrations for manufacturing floors.
  • General service availability is a notable concern, with three serious outages and system malfunctions occurring in 2025 alone.
  • Custom scripts rely on specialized proprietary languages, making in-house maintenance difficult.

Evaluation Summary:
This platform offers powerful global business consolidation and an all-in-one corporate record database. Even so, companies must be ready for sharp pricing shifts at renewal and evaluate alternative integrations if they run complex manufacturing or shop-floor operations.

Odoo Enterprise

• Pros:

• • Modular system architecture allows businesses to install only the apps they currently need.
  • Modern user interface ensures a fast learning curve for internal teams.
  • Large international open-source community provides a massive library of community apps.
• Cons:
  • Heavy customization paths often break or encounter validation errors during major version upgrades.
  • The official documentation lacks depth regarding complex multi-currency accounting rules.
  • Connecting independent modules frequently leads to data validation mismatches.
  • Relies heavily on third-party community extensions to handle region-specific tax reporting.

Evaluation Summary:
A flexible, app-based ecosystem lets organizations build out their digital platform step-by-step. It provides a solid baseline for companies with standard workflows, provided they carefully manage app dependencies to avoid issues during system upgrades.

Also Read: Top 5 best HRMS software for large enterprise with multiple workplaces in Singapore

Regional Operational Needs

Running an enterprise within this specific financial hub brings unique operational challenges compared to Western markets. Software cannot just handle basic book-keeping; it must natively adapt to the high-velocity trade environments of the region.

• Advanced Cross-Border Trade Support:Platforms must handle complex multi-currency transactions and real-time exchange adjustments automatically, keeping regional supply chains moving without manual recalculations.
• Strict Digital Compliance:Systems must align smoothly with strict local tax structures and electronic filing mandates, ensuring clean audit trails for regional authorities.
• Agile No-Code Workflows:Fast-moving business environments require teams to modify operational modules instantly. Solutions like Multiable ERP let users adjust layouts and workflows on the fly without waiting for lengthy IT development cycles.

Static vs. Dynamic Model Context Protocol

Understanding the difference between static and dynamic Model Context Protocol (MCP) is vital when planning an AI-driven enterprise strategy:

• Static MCP:This configuration only supports the built-in, out-of-the-box standard functions of an enterprise system. If a business creates a custom fields array or sets up a unique operational workflow, the background AI agent remains blind to those areas. It cannot access or read the customized data points without manually writing new integration code.
• Dynamic MCP:This advanced protocol automatically maps and exposes the entire active system database to AI models. It covers standard modules as well as custom functions configured or developed by the end-user. As a result, autonomous AI agents can instantly understand, query, and run automated workflows across the company’s entire unique operational setup.

System Selection Risks to Watch in 2026

Meanwhile, according to a study by Everfair Management, when evaluating modern enterprise software architectures, business owners need to focus on technical challenges that have emerged alongside recent AI developments:

1. Ecosystem Infrastructure Boundaries:Avoid purchasing platforms that are completely bound to the Windows Server ecosystem. Because the vast majority of popular LLMs and cutting-edge agentic AI tools run natively on Linux, enterprise platforms that cannot run on Linux risk becoming obsolete in the near future.
2. The Shift Toward Regional Innovation:While AI tools in Western markets dominated early conversations, Asian software providers haSve rapidly closed the gap. Regional platforms now frequently deliver a significantly higher ROI than traditional enterprise software names from the US or EU by offering native, cost-effective automation tailored to local workflows.
3. The Stability of Direct Vendor Procurement:Purchase directly from the core software developer instead of relying on a consultation partner or third-party reseller. The long-term service quality and business sustainability of a reseller are inherently weaker and more volatile than dealing directly with the company that builds and maintains the platform.
4. Built-In MCP Requirements:Do not purchase a system that lacks native, built-in MCP support. Without this layer, attempting to securely route corporate operational data out to modern AI tools results in massive integration costs and fragile data connections.
5. Native Custom API Generation:Ensure the platform includes an automated API generator for user-configured modules. Without this feature, adding a custom workflow requires a slow, expensive development loop: writing the custom function, manually building a corresponding API, and hard-coding that data into a rigid interface.

Why We Write this Article?

This piece is authored by Sam Cheong, the Principal Consultant at Synchro RKK Sdn Bhd and one of Malaysia’s most respected business software authorities. Driven by a passion for complex problem-solving, Sam fell in love with Enterprise Resource Planning (ERP) architecture early in his career. Following a proven track record of high-impact deployments, he successfully acquired the ERP business unit from SRKK to found Synchro ERP. Today, he leverages his deep technical expertise and strategic vision to help organizations streamline operations, scale infrastructure, and navigate digital transformation. Witnessing the rising wave of compromised implementations in the region, Sam shares these insights to guide enterprises away from costly deployment pitfalls.

—

Want updates like this delivered directly? Join our WhatsApp channel and stay in the loop.

We can share your story at e27 too! Engage the Southeast Asian tech ecosystem by bringing your story to the world. You can reach out to us here to get started.

Featured Image Credit: Canva Images

The post Addressing the Digital Shift – AI Accounting ERP Trend in Singapore appeared first on e27.

Ask anyone to name Southeast Asia’s most successful startups and you will hear the same names. Grab. Sea. GoTo. Each is a consumer company. Each won by reaching enormous numbers of users, burning enormous amounts of capital, and surviving long enough to consolidate a market.

This is the story the region tells about itself. It is also the reason Southeast Asia is structurally underbuilt in the categories that, almost everywhere else in the world, produce the most durable companies.

The boring layer is where durable value compounds

Globally, the companies that compound value over decades are rarely the exciting consumer names. They are the unglamorous infrastructure businesses that sit underneath the economy: supply chain software, vertical applications for traditional industries, payments rails, logistics tooling, compliance systems. These companies are boring to talk about. They are also extraordinarily hard to displace once they are embedded, which is exactly why they compound.

Southeast Asia has underweighted this layer for a decade. The capital chased consumer scale because consumer scale produced headlines, and the founder pipeline followed the capital. The result is a region with world-class consumer platforms sitting on top of a thin and underbuilt layer of business infrastructure.

The data shows both the gap and the turn.

Southeast Asia’s B2B digital commerce market passed US$90 billion in 2024 and is heading toward US$130 billion by 2026. Average SaaS spending per employee in the region rose from US$3.79 in 2020 to US$13.47 in 2025, a 2.5-fold increase in five years. The share of software and services in emerging industry investment jumped from 28 per cent in the second half of 2023 to 50 per cent in the first half of 2024. On the physical side, McKinsey estimates a roughly US$60 billion gap between existing or announced supply chain infrastructure investment and what the region’s future trade flows will require.

These are not the numbers of a mature market. They are the numbers of a market that has barely started building its business infrastructure, growing quickly from a very low base.

Also Read: What Southeast Asia’s edutech startups can learn from AI’s limits in education

Why the underbuild persists

Three forces keep the boring layer underfunded.

The first is capital habit. A decade of consumer outcomes trained the region’s investors to look for winner-take-all scale stories. Business infrastructure does not look like that early on. It grows through slow, defensible adoption inside individual industries, not through viral user curves. To an investor pattern-matching on the last cycle, a vertical software company solving a specific operational problem for mid-sized manufacturers looks small. It is not small. It is early.

The second is founder training. Founders in the region have been taught, implicitly, that the prize is consumer scale. The most ambitious technical talent gravitates toward consumer problems because that is where the celebrated outcomes have been. The boring problems, the ones embedded deep in logistics, procurement, financial operations, and regulatory compliance, attract less talent than their economic importance warrants.

The third is visibility. A consumer app is legible to everyone. A supply chain traceability platform serving Vietnamese manufacturers under pressure to meet global compliance standards is invisible to most observers, even though it may be solving a more durable and defensible problem than most consumer apps ever will.

The categories that are most underbuilt

For founders and investors willing to look at the boring layer, several categories in Southeast Asia are conspicuously underbuilt relative to the size of the problem they address.

Supply chain and logistics tooling, where the region’s emergence as a manufacturing alternative to China is creating demand far faster than software is being built to serve it. SME financial software, where tens of millions of small and mid-sized businesses still run on spreadsheets and manual processes. Vertical SaaS for traditional industries, where deep workflow integration creates the network effects and switching costs that make companies durable. Compliance and regulatory technology, where fragmented rules across ten markets create a problem that is painful, recurring, and exactly the kind of thing businesses pay for indefinitely. B2B payments and financing infrastructure, where cross-border trade is growing but the rails underneath it remain thin.

None of these will produce the next viral consumer story. All of them can produce companies that are still compounding in twenty years.

Also Read: The capital drought: Over 7,500 SEA startups extinguished since 2020

What this means for capital allocation

The investors who recognise this early have an advantage that the consumer cycle no longer offers. Vertical software companies in the region are already attracting stronger conviction precisely because they show what consumer companies often could not: sustainable unit economics and high retention. Sector-focused funds are increasingly treating these as defensible plays that can dominate a niche before global incumbents arrive.

The thesis is straightforward. The next generation of durable Southeast Asian companies will disproportionately be built in the boring categories, because that is where the combination of real demand, defensibility, and underbuilt supply is strongest. Capital that continues to crowd into the exciting categories will compete for diminishing returns. Capital that moves into the boring categories will be early to the layer where durable value is actually created.

A region’s most valuable companies are not always its most visible ones. Southeast Asia spent a decade building the visible layer. The durable companies of the next decade are being built underneath it, in the categories most people find too boring to watch.

That is exactly why they are worth watching.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post Why Southeast Asia is underbuilt in the categories that produce its most durable companies appeared first on e27.

MDI Ventures, the corporate venture capital arm of Indonesia’s TelkomGroup, is narrowing its Southeast Asia investment focus around enterprise AI, AI infrastructure, governance software, real-world asset tokenisation, institutional custody, and regulated digital asset rails.

The Jakarta-headquartered firm said its updated thesis reflects a shift in the region’s technology cycle. After more than a decade in which consumer internet, e-commerce, ride-hailing, and digital payments defined Southeast Asia’s startup story, MDI believes the next phase will be shaped by the infrastructure and software needed to make AI and digital assets usable at enterprise scale.

That means less focus on broad digitisation as a theme in itself and more on the systems behind it: compute, data governance, cybersecurity, compliance, workflow automation, payment settlement, custody, and tokenisation.

Also Read: Why Singapore, Indonesia, and Vietnam are losing the AI race they think they are winning

For MDI, the strategy is also tied closely to TelkomGroup’s domestic reach. The firm plans to source more companies from Southeast Asia, beginning with Singapore, and help them enter Indonesia through TelkomGroup’s enterprise relationships, digital infrastructure, market channels, and strategic partners.

“Southeast Asia’s next wave of technology growth will be led by companies that solve real infrastructure and enterprise problems,” said Shannon Lee Chaluangco, Investment Director at MDI Ventures Singapore. “Our role is to provide the strongest regional founders with Indonesia’s market scale, TelkomGroup’s network, and the strategic resources needed to build durable companies across AI, blockchain, and digital assets.”

Why Singapore-to-Indonesia matters

The Singapore-Indonesia corridor has long been one of Southeast Asia’s most important startup routes. Singapore offers access to capital, regional headquarters, regulators, multinational customers, and technical talent. Indonesia offers market scale: more than 270 million people, a large enterprise base, and a fast-growing digital economy still unevenly served by modern software.

MDI’s updated thesis sits directly on that bridge. Singapore remains Southeast Asia’s leading AI funding hub, capturing around 57 per cent of the region’s AI funding over the 12 months to June 2025. During that period, 495 AI startups in the city-state raised US$1.31 billion.

For many of these companies, however, regional expansion is difficult. Indonesia is not a plug-and-play market. Enterprise sales cycles can be relationship-driven, regulatory interpretation varies by sector, and localisation often goes beyond language. Founders must understand local data rules, procurement behaviour, payment preferences, and how large incumbents actually buy technology.

Also Read: Indonesia’s AI hiring gap is real, just not 28×

This is where corporate venture capital can be more useful than capital alone. MDI is betting that its link to TelkomGroup can give founders a faster route into live commercial deployments, especially in sectors where trust, distribution, and regulatory familiarity matter.

AI moves from pilots to production

The firm’s timing reflects a wider reallocation of venture capital. In 2025, AI companies accounted for 61 per cent of global venture capital investment, representing US$258.7 billion. In Southeast Asia, more than US$2.3 billion flowed into roughly 680 to 700 active AI startups over the 12 months to June 2025, equal to about 32 per cent of all private funding in the region.

That resilience stands out because the broader funding environment has been much weaker. Southeast Asian private funding fell to a six-year low of US$1.85 billion across 229 deals in the first half of 2025. AI, in other words, has become one of the few categories still able to attract strategic capital at scale.

Indonesia also presents a large deployment opportunity. MDI’s thesis work points to roughly 2,400 AI and deep tech startups in the country, while 92 per cent of knowledge workers already use AI, compared with a global average of 75 per cent. Yet among 336 Indonesian AI companies tracked by the firm, only 66 have raised institutional funding.

That gap suggests a market where usage is running ahead of venture backing. MDI sees opportunities in “Bahasa-first” AI applications, workflow automation, customer engagement tools, data governance, compliance software, and sector-specific AI systems for industries such as telecommunications, finance, logistics, and public services.

The infrastructure layer is just as important. Indonesia’s data centre pipeline is projected to reach around 1,519MW by 2028, while cloud adoption across Southeast Asia is growing at about 20 per cent annually. As companies move from AI experiments to production systems, they will need more than chat interfaces. They will need secure deployment, model monitoring, integration with legacy systems, and clear governance over how data is used.

Digital assets beyond speculation

MDI’s second major focus is blockchain and digital assets, but not in the retail trading sense that dominated the previous crypto cycle. The firm is looking at institutional infrastructure: custody, tokenisation, payments, settlement, compliance, and regulated digital asset services.

The market case is large, though still early. The potential market for tokenised real-world assets is projected to reach US$30.1 trillion within the next decade. More than 70 per cent of Southeast Asia’s internet users now use digital financial services, and digital payment gross transaction value in the region is expected to reach US$359 billion by 2025.

Adoption signals are also strong. Asia accounts for six of the world’s top ten countries in crypto adoption. Indonesia recorded about US$3.1 billion in crypto transaction volume in May 2024, up 506 per cent year-on-year. Vietnam’s annual digital asset transaction volumes exceed US$220 billion, while Singapore has built an ecosystem of more than 300 blockchain companies.

Also Read: SBI buys majority stake in Coinhako to deepen Singapore digital asset push

Still, the next phase will likely depend less on retail enthusiasm and more on whether regulated institutions can safely use blockchain-based rails. Tokenised bonds, fund units, carbon credits, invoices, and other real-world assets require custody, identity checks, settlement processes, and legal clarity. That makes compliance a core product feature, not an afterthought.

“Web3 in Southeast Asia is moving from speculation to infrastructure,” said Alvin Evander, VP of Strategy at MDI Ventures. “As tokenisation, institutional custody and regulated digital asset rails move from pilots into real-world use, the winners will be founders who build for trust and compliance from day one.”

MDI has been raising its profile in this space. It participated as an institutional partner at Indonesia Blockchain Week 2026 in Jakarta and co-hosted “Tokenize Indonesia – Brunch by the Beach” at Coinfest Asia 2026 in Bali, bringing together regulators, financial institutions, and industry players. The firm also plans to meet founders and ecosystem partners at TOKEN2049 Singapore.

The competitive field

MDI is not alone in chasing the region’s next infrastructure cycle. Corporate and strategic investors such as Singtel Innov8, SCB 10X, Krungsri Finnovate, and Vertex Ventures have also backed startups across enterprise software, fintech, AI, and blockchain-linked infrastructure. In Indonesia, firms such as East Ventures, AC Ventures, Alpha JWC Ventures, and BRI Ventures remain active across technology sectors. MDI’s differentiator is its connection to TelkomGroup, but that advantage will matter only if it translates into real contracts, integrations, and distribution for portfolio companies.

Also Read: Four VC executives. Zero personal gain. Three years in prison

For founders, the updated thesis signals where strategic capital may be heading next. Southeast Asia’s startup market no longer rewards growth stories as generously as it did during the zero-interest-rate boom. Investors are asking harder questions about revenue quality, defensibility, regulation, and enterprise adoption.

MDI’s bet is that the next durable companies in the region will not simply digitise existing behaviour. They will build the rails that make AI and digital assets safe, compliant, and commercially useful in large markets such as Indonesia.

The post MDI Ventures sharpens SEA thesis around AI, blockchain, digital asset infra appeared first on e27.

For many large companies in Southeast Asia, the challenge with artificial intelligence (AI) is no longer access. The models are available, the cloud infrastructure is ready, and boards have approved AI experimentation.

The harder problem is turning that experimentation into systems that work inside heavily regulated businesses without breaking compliance, confusing staff or creating new operational risks.

That is the gap Singapore-headquartered OneByZero, or OBZ, is trying to fill. The company has raised US$20 million in Series A funding led by Jungle Ventures, marking its first external fundraise after three years of operating with enterprise customers across the region.

Also Read: Singapore’s AI dividend will depend on what happens after the pilot phase

OBZ describes itself as a “frontier AI deployment company”, a phrase that reflects a broader shift in the AI market. Rather than selling a standalone software tool, the company places engineering teams inside large organisations to identify where AI can create value, build the technology into existing workflows, and keep improving it after deployment.

Its focus is on regulated and operationally complex sectors such as financial services, telecommunications and retail. These are industries where AI pilots are common, but production deployments tend to move slowly because systems are old, data is fragmented, and mistakes can carry reputational or regulatory consequences.

“Every enterprise can now get access to powerful AI. The value comes when it runs inside the systems that run the business, under the company’s own rules, and gets better every week,” said Niket Vaidya, CEO and co-founder of OneByZero. “That is the work we do every day, and this funding lets us do more of it, in more markets, and more industries.”

The deployment problem

OBZ’s fundraise lands at a time when enterprises across Asia Pacific are under pressure to prove that AI spending can produce measurable returns. Since generative AI tools went mainstream, companies have run pilots in customer service, internal knowledge management, software development, fraud monitoring, marketing and document processing. But many of these efforts remain stuck at the proof-of-concept stage.

The reasons are familiar to enterprise technology teams. AI systems need access to business data, but that data often sits across multiple legacy systems. Customer-facing use cases require strict guardrails. Regulated industries need audit trails. Staff need to know when AI can act on its own and when a human must intervene.

OneByZero says its approach is built around these realities. Its forward-deployed engineers work alongside customer teams to identify high-value use cases, build and deploy systems, then stay involved to run and improve them. It also operates local teams in each market, which it says helps customers deal with language, regulatory and systems differences.

The company claims its deployments have helped enterprises automate more than 90 per cent of some customer-facing interactions, generate measurable cost savings, and accelerate complex data modernisation projects by 50 per cent. It did not disclose customer names or revenue figures, but said it has more than doubled revenue annually over the past three years.

Also Read: Language was never the problem: Inside SEA’s real AI adoption gap

For Southeast Asia, this kind of work is likely to become more important. The region’s largest banks, telcos and retailers often operate across several markets, each with its own rules, languages and consumer behaviours. An AI system that works in Singapore may need significant adjustment before it can be used in Indonesia, Thailand, Vietnam or the Philippines. That makes deployment expertise, not just model access, a meaningful differentiator.

NEO and the rise of AI coworkers

At the centre of OBZ’s offering is NEO, its AI deployment platform. The company describes NEO as the control layer behind enterprise AI workforces, combining reusable components with governance features such as records of every agent action.

In simple terms, NEO is meant to help companies define what AI agents are allowed to do, where human approval is required, and how those actions are tracked. OneByZero calls these agents “AI Coworkers” — governed AI systems with defined roles and controls. The idea is that AI handles high-volume, repetitive work, while humans retain judgement and accountability.

“We have built NEO so companies can define what AI is allowed to do, where human judgement is required and how every system is governed,” said Vibhore Kumar, PhD, CTO and co-founder of OneByZero. “That foundation lets them introduce more capable AI without losing control.”

The governance point is crucial. As AI agents become more capable, enterprises are moving beyond chatbots and copilots into systems that can take actions: updating records, generating responses, flagging transactions, routing cases or initiating workflow steps. In sectors such as banking, telecoms and healthcare, companies will need clear limits on what those agents can do.

This is also where smaller or open-weight language models may play a role. OBZ said part of the new funding will go into helping enterprises build their own AI on open-weight and small language models. For regulated companies, smaller models can sometimes be easier to customise, control and run in specific environments than general-purpose large models. They may also cut costs for repetitive tasks where a frontier model is not required.

Expansion across Asia Pacific and Japan

OBZ currently operates in nine markets: Australia, India, Indonesia, Malaysia, the Philippines, Singapore, Thailand, the US and Vietnam. It plans to use the Series A funding to grow its forward-deployed engineering teams and deepen customer relationships in these markets.

It also plans to enter Japan, one of Asia’s largest enterprise technology markets. Japan is attractive because of its large corporate base, advanced manufacturing and financial sectors, and urgent need for productivity tools as the country grapples with labour shortages and an ageing population. But it is also a difficult market to crack, requiring local relationships, language capability and long enterprise sales cycles.

Also Read: fileAI expands in Japan with new backing from SMBC and Singtel Innov8

Beyond financial services and telecommunications, OBZ wants to expand into conglomerates, healthcare and the public sector. These sectors share many traits with its current customers: complex operations, large volumes of data, and a need for careful governance.

Jungle Ventures’s investment reflects growing investor interest in companies that can turn AI adoption into practical enterprise outcomes rather than simply building model wrappers. “OneByZero stood out because it is already doing what much of the enterprise AI market is still talking about,” said Yash Sankrityayan, Managing Partner at Jungle Ventures. “The team has spent three years deploying AI inside large, complex organisations and delivering measurable results.”

A crowded but evolving field

OBZ is not alone in chasing the enterprise AI deployment opportunity. Globally, it competes for attention and budgets with companies such as Palantir, C3.ai and DataRobot, and with large consulting firms including Accenture, Deloitte, IBM and Capgemini, all of which help enterprises operationalise AI. In Asia, systems integrators, cloud partners and boutique AI consultancies are also moving into the same space.

Its differentiation will depend on whether it can prove that its forward-deployed model and NEO platform deliver repeatable results across markets and industries. That is not easy. Service-heavy AI deployment businesses can be difficult to scale, while platform businesses need standardisation. OBZ is trying to sit between the two: using hands-on engineering to solve messy enterprise problems, then turning what it learns into reusable templates and governance tools.

Also Read: Why Southeast Asian enterprises need AI governance before scaling generative AI

That middle ground could be valuable in Southeast Asia, where companies often need local execution as much as software. But it also means OBZ will have to keep hiring strong technical teams in multiple markets while maintaining quality and consistency.

The broader market direction is clear. Enterprises are moving from asking what AI can do to asking where it can be trusted to act. OBZ’s bet is that the next wave of AI adoption will be won not by the companies with the flashiest models, but by those that can make AI work inside the unglamorous systems that actually run large businesses.

The post OneByZero raises US$20M Series A to help enterprises move AI from pilots to production appeared first on e27.

Singapore’s electric vehicle (EV) market is no longer waiting for a breakthrough moment. In 2026, the shift is already visible on the road, in dealerships, and across public car parks where chargers have become a more common part of the urban landscape. But the next stage of growth may be harder than the first.

EV sales in Singapore are forecast to rise 34.3 per cent in 2026 to 34,940 units, according to BMI Country Risk and Industry Research, a Fitch Group unit. That would lift EVs to 54.6 per cent of total vehicle sales this year, underlining how quickly electric mobility has moved from early adopter territory into the mainstream.

Also Read: VinFast’s ride-hailing arm Green SM takes on Jakarta’s ojek economy with e-scooters

The wider fleet is also changing. BMI expects Singapore’s EV fleet to reach 92,705 units in 2026, or 10.7 per cent of the country’s total vehicle fleet. By 2035, the research house expects EVs to account for 44 per cent of all vehicles in Singapore, with EV sales making up 85 per cent of total vehicle sales.

Those numbers point to a market that is accelerating fast. Yet they also highlight a more complicated reality: Singapore’s EV transition is becoming increasingly dependent on policy design, charging availability, fleet electrification and the economics of ownership in a country that does not want too many cars in the first place.

A fast-growing market, but not a typical one

Singapore’s EV adoption has gathered pace rapidly in 2026. As of June, EVs accounted for 70 per cent of new vehicle registrations. In the first half of the year, EV adoption reached a record 62.4 per cent of new car registrations, up 110 per cent year on year.

The first quarter marked an important turning point. EVs made up 57.6 per cent of new car registrations, or 7,679 vehicles, the first time electric cars outsold internal combustion engine and hybrid vehicles combined, according to BMI. In June alone, EVs accounted for 4,791 new vehicle registrations.

This makes Singapore one of Southeast Asia’s most advanced EV markets by adoption share, even if its absolute vehicle volumes remain small compared with larger neighbours such as Indonesia, Thailand or Malaysia. The city-state’s high-income consumer base, compact geography and strong regulatory capacity make it well suited to EV rollout. Range anxiety is less severe than in large countries, while dense housing and public car parks allow charging infrastructure to be planned with more central coordination.

But Singapore is also unlike most car markets. Vehicle ownership is deliberately constrained through the Certificate of Entitlement system, high registration costs and a long-standing “car-lite” strategy that prioritises public transport. That means EV adoption can grow quickly as a share of new sales, even while the total number of private cars remains tightly managed.

BMI expects passenger EV sales growth to moderate over the longer term, with an average annual growth rate of 5.3 per cent between 2026 and 2035. The constraint is not only consumer demand. It is the structure of Singapore’s transport policy.

Incentives are doing heavy lifting

Government incentives remain one of the most important drivers of adoption. Singapore’s EV Early Adoption Incentive, which runs until December 2026, provides new electric cars and taxis with a 45 per cent rebate on the Additional Registration Fee, capped at US$5,864 in 2026.

The Enhanced Vehicular Emissions Scheme also supports demand, offering rebates of up to about US$17,592 for qualifying cars in 2026 and around US$15,637 in 2027. These incentives matter because Singapore is one of the world’s most expensive places to own a car. Even a cheaper EV can carry a high upfront cost once taxes, fees and the Certificate of Entitlement are included.

Also Read: Thailand’s mobility future will be decided by data, not just vehicles

At the same time, policy is becoming more exacting. From 2026, the Vehicular Emissions Scheme band thresholds were recalibrated to align with stricter real-world measurements under the Worldwide Harmonised Light Vehicles Test Procedure. BMI said this will affect the rebates and surcharges applied to new vehicle registrations.

The recalibration is important because Singapore is not simply trying to replace petrol cars with electric cars. It is trying to reduce transport emissions while maintaining tight control over road usage. In that context, incentives must encourage cleaner vehicles without undermining the broader push towards public transport.

This is where the country’s EV story differs from other Southeast Asian markets. Thailand and Indonesia are using EV policy partly to build manufacturing supply chains. Singapore’s focus is demand-side adoption, infrastructure readiness and emissions reduction within a dense urban transport system.

Chinese brands change the price equation

Another major force behind Singapore’s EV momentum is the arrival of more affordable models, particularly from Chinese manufacturers. BYD has become the standout example.

According to BMI, BYD accounted for 25 per cent of Singapore’s passenger vehicle market in the first half of 2026, compared with 12.5 per cent for Toyota. In the first quarter, BYD represented 24.3 per cent of new vehicle registrations with 3,239 units, making it the market leader.

This reflects a wider regional pattern. Chinese EV makers have been expanding aggressively across Southeast Asia, bringing lower-cost models, battery expertise and faster product cycles. In markets where EVs were once associated mainly with premium brands, Chinese manufacturers have helped shift the conversation towards affordability and practicality.

For Singapore buyers, the impact is amplified by the cost of ownership. Any reduction in the vehicle’s base price can make a meaningful difference once rebates and registration costs are applied. But it also increases competitive pressure on Japanese, Korean and European automakers that have long been familiar names in Singapore.

Charging network becomes the next test

As EVs move into the mainstream, charging infrastructure becomes more than a convenience issue. It becomes a confidence issue.

Singapore had around 30,500 EV charging points as of March 2026, nearly double the roughly 15,300 recorded in November 2024. The government aims to install 60,000 charging points by 2030 under the Singapore Green Plan 2030, including 40,000 in public car parks and 20,000 at private properties such as residential developments and offices.

That target matters because many Singaporeans live in high-rise public or private housing and do not have access to private garages. Unlike landed home owners in other markets, they depend heavily on shared charging infrastructure in car parks, workplaces and commercial areas.

Also Read: Tesla establishes Vietnam subsidiary as EV rivalry with VinFast looms

The charging market is also getting more competitive. BMI said there were 36 EV charging operators in Singapore as of July 2026, although it expects consolidation as companies compete in a crowded field. In June, SP Mobility completed its acquisition of ChargEco, integrating more than 1,000 public charging points and becoming the operator of Singapore’s largest EV charging network.

Regulation is also evolving. In March 2026, Singapore raised its national EV charging standard from Technical Reference 25 to SS 722. The new standard includes requirements for smart-grid integration, electrical safety, battery-swapping protocols and updated direct-current fast-charging specifications.

Fleets may drive the next wave

Commercial vehicles could become an important part of the next growth phase. BMI forecasts commercial EV sales to rise 18.5 per cent in 2026 to 2,652 units, after growing 63.4 per cent in 2025. From 2026 to 2035, commercial EV sales are expected to grow at an average annual rate of 8.4 per cent.

The drivers are clear: pressure on businesses to decarbonise supply chains, government and municipal fleet electrification, and better charging infrastructure. Incentives also help. The Commercial Vehicles Emissions Scheme, which runs until March 2027, provides incentives of up to about US$15,637 for the least-polluting commercial vehicles and penalties of up to around US$11,728 for the most polluting.

For heavy vehicles, the Heavy Vehicle Zero Emissions Scheme supports businesses registering new zero-tailpipe-emission heavy goods vehicles and buses. The incentive was reduced to about US$11,728 from September 2026, except for vehicles with a maximum laden weight above 7,000kg, after strong take-up and a narrowing cost gap with internal combustion models.

BMI expects Singapore’s electric heavy commercial vehicle segment to remain small at about 212 units in 2026, but forecasts it to reach 1,131 units by 2035. Bus fleet electrification should also provide medium-term support.

Still, not all consumer sentiment is moving in one direction. A 2026 study cited by BMI found that 32 per cent of respondents planned to buy an internal combustion engine vehicle over the next two years, up from 26 per cent in 2024. Concerns over charging availability and hidden costs were among the reasons.

Also Read: Waymo’s Singapore entry raises the stakes for autonomous mobility in Asia

That is the paradox of Singapore’s EV transition. Adoption is rising quickly, but future growth will not come from a simple expansion of private car ownership. Public transport remains central: MRT, light rail and bus networks recorded an average of 7.2 million daily rides in 2023, and the government wants at least 80 per cent of households to be within a 10-minute walk of a train station by 2030.

Singapore’s EV market is therefore entering a more mature phase. The easy story is that electric cars are winning. The harder story is what comes next: keeping incentives calibrated, making charging reliable, electrifying fleets, and ensuring EV adoption supports, rather than competes with, the country’s broader car-lite future.

The post EVs to make up over half of Singapore vehicle sales in 2026, BMI says appeared first on e27.

For most artificial intelligence companies, Singapore is either a regional sales base or a policy-friendly testbed. For Zero Gravity, better known as 0G, the city-state is now becoming something more central: its global headquarters.

The AI infrastructure company said it will invest about US$15.4 million in Singapore over the next five years, using the country as a base to grow its operations, research partnerships and developer ecosystem.

The announcement was made at 0G Dev Day Singapore 2026 at the National Gallery Singapore, where close to 3,000 participants registered, including developers, researchers, businesses, investors and members of the wider technology community.

Also Read: SEA’s venture capital shifts from mega-rounds to AI and SaaS

The move places 0G in a country that has been working to position itself as a trusted AI hub for Southeast Asia and beyond. Singapore’s National AI Strategy 2.0 has put emphasis on talent, compute access, governance and practical adoption, while the city’s regulators have tried to balance innovation with safety. For companies building AI infrastructure, that mix is attractive: Singapore offers proximity to regional markets, deep pools of capital and a government keen to shape global AI standards.

Senior Minister of State for Digital Development and Information and Health Tan Kiat How, who delivered the keynote at the event, framed Singapore as a place where AI products can be deployed, adapted and scaled. That message fits neatly with 0G’s own pitch: build in Singapore, then export globally.

Michael Heinrich, CEO and co-founder of 0G, said the company’s commitment is “not simply about opening an office” but about “research, talent, innovation and long-term investment”. He added: “We see Singapore not as a final destination, but really as a launchpad for the rest of the world.”

A headquarters play, not just an office opening

Of the planned US$15.4 million investment, around US$7.7 million will go towards establishing Singapore as 0G’s global headquarters, building operations, growing the team and developing its ecosystem.

The company has already committed about US$3.9 million to a four-year research collaboration with Nanyang Technological University (NTU), focused on decentralised AI and blockchain-based infrastructure. It plans to commit another US$3.1 million to research collaborations with other universities and academic partners. A further US$770,000 will support AI and badminton research with NTU and YB Badminton Academy.

0G’s first physical office globally is also in Singapore, at Stamford Place. The location is symbolically convenient: the building formerly housed Singapore’s National Heritage Board, a detail Heinrich has used to position the company as one building future infrastructure from a site linked to the country’s past.

Also Read: Singapore’s Psalion raises US$50M fund for Web3’s next practical phase

At a practical level, the decision reflects a broader pattern in Southeast Asia’s tech ecosystem. Singapore continues to attract regional headquarters for startups, fintech firms, AI companies and crypto-related infrastructure players, even when their markets, users and developer communities are spread across Indonesia, Vietnam, the Philippines, India and beyond. Its strengths lie less in domestic market size and more in regulation, talent mobility, institutional partnerships and investor access.

Taking decentralised AI from lab to court

The more unusual part of 0G’s announcement is not the headquarters plan, but where some of its research is going next: badminton.

At Dev Day, Zero Gravity Labs and NTU formalised the next phase of their existing US$3.9 million research collaboration, expanding work in decentralised AI towards real-world applications. Sports will be an initial focus, with badminton as the starting point.

The partnership, established in 2025, has been working on decentralised AI training, model alignment and verifiable AI systems. Researchers from 0G and NTU have had three collaborative research papers accepted at NeurIPS and one at ICML, two of the most closely watched machine learning conferences globally.

Badminton is not a random choice. In Southeast Asia, the sport has deep cultural and competitive significance, particularly in Indonesia, Malaysia, Thailand and Singapore. Singapore’s Loh Kean Yew, a former world champion, has also helped raise the sport’s profile locally, while his Asian Games silver added further momentum.

For AI researchers, badminton offers dense and fast-moving data. A single rally can involve rapid changes in body position, shuttle trajectory, footwork, shot selection, fatigue and tactical decision-making. Much of that is hard to capture in one clean dataset.

Working with NTU and YB Badminton Academy, 0G plans to explore a badminton foundation model that combines data from players, movement, video, training and competition. Foundation models are AI systems trained on broad datasets and then adapted for specific tasks. In this case, possible uses include helping coaches spot movement patterns that are difficult to see in real time, identifying training loads that may increase injury risk, and supporting young players with more personalised feedback.

“Badminton brings together frontier AI research, real-world data and a very human outcome: helping people perform better and stay healthier over time,” Heinrich said.

Also Read: Malaysia’s sovereign AI bet: Local context becomes the next startup moat

0G Singapore also signed a three-year memorandum of understanding with YB Events and Sports for Project New Frontiers: Bringing Badminton to New Horizons. Under the partnership, 0G Singapore will commit about US$204,000 annually for three years to support youth badminton development, international competition and the exploration of AI and emerging technologies in sport.

The trust layer question

Beyond sport, 0G is pitching itself as part of the infrastructure stack needed for safer AI adoption. The company describes its mission as building the “trust layer for AI”, with decentralised infrastructure designed to make AI more private, verifiable, sovereign and open.

At Dev Day, it showcased 0G Private Computer, a product aimed at bringing privacy and verification to sensitive AI workloads. Heinrich said the system is designed so that sensitive workloads can remain private, including from 0G itself. The broader idea is to give organisations more confidence over how data is handled and how AI workloads are processed, especially when proprietary information is involved.

The company also introduced BestCEO, a ready-to-use AI solution aimed at making advanced AI capabilities accessible to businesses, including small and medium-sized enterprises, without requiring them to build AI infrastructure or applications from scratch.

This is a relevant angle for Southeast Asia, where SMEs make up the backbone of most economies but often lack the technical teams or budgets to adopt AI meaningfully. The region’s AI opportunity is not only about large enterprises deploying copilots, but also about whether smaller companies can use the technology without handing over sensitive data or depending entirely on foreign platforms.

Rivals in a crowded AI infrastructure race

0G is entering a crowded and fast-shifting market. In decentralised AI and compute, it sits alongside global projects such as Bittensor, Gensyn, Akash Network, Render Network and Aethir, all of which are trying to rethink how AI models, compute resources and data networks are coordinated. It also competes indirectly with hyperscale cloud providers such as Amazon Web Services, Microsoft Azure and Google Cloud, which remain the default infrastructure providers for most AI companies. The key question for 0G is whether developers and enterprises will see decentralisation and verifiability as must-have infrastructure, rather than a technically elegant alternative to existing cloud systems.

Community as a test of accessibility

0G is also trying to position AI access as a community issue, not just an enterprise problem. At its first community AI workshop in Singapore, 70 residents, including many seniors, learned about AI and built websites and applications on 0G over two days. One senior participant built a website to help people understand which health screenings they should consider at different ages, including reminders for women about mammograms.

Also Read: Crypto is dead? Apparently not, says Y Combinator – Blockchain is still worth building

That example is small, but it points to the broader challenge around AI adoption in the region. For AI to move beyond boardrooms and developer conferences, it must become useful to ordinary users, schools, coaches, seniors and small businesses.

As 0G plants its headquarters in Singapore, the company is making a familiar but ambitious bet: that the city-state can serve as a neutral, trusted base for technologies meant to travel far beyond its borders. The harder test will be whether its infrastructure can move as smoothly from research papers to badminton courts, SMEs and real-world AI systems.

The post 0G picks Singapore as global base, expands AI research into badminton appeared first on e27.

The post The agent that lied: what GPT-6.1 Astra’s cancellation means for Southeast Asia appeared first on e27.

Southeast Asia has become one of the world’s most competitive regions for investment.

Global companies are diversifying supply chains beyond China. Governments are offering tax incentives, industrial land and digital-economy programmes. New capital is flowing into manufacturing, data centres, semiconductors and technology companies.

Yet investors evaluating the region are looking beyond growth rates and startup potential.

They are also evaluating governments.

Will tax rules remain stable? Can foreign investors retain control of their companies? Will data be allowed to move across borders? Will incentives survive the next budget? Could a new regulation make an existing business model unviable?

These questions shape investment decisions more than many founders realise.

Capital is attracted by growth. It stays where the rules remain understandable.

Policy uncertainty has a price

Businesses do not always need low taxes or light regulation. They need rules they can plan around.

A company can model a 20 per cent corporate tax rate. It can adapt to foreign ownership restrictions. It can comply with strict data-protection laws.

What is much harder to manage is constant change.

If a licence normally takes nine months, a company can include that delay in its plans. If approval might take three months, two years or never arrive, the risk becomes harder to calculate.

The result is often predictable. Investors delay decisions, reduce the size of projects or choose another market.

Regulatory uncertainty effectively becomes an additional cost. Investors demand higher returns to compensate for it. They commit less capital and plan over shorter periods.

Startups are especially exposed.

Large corporations can hire legal teams, advisers and compliance specialists. Early-stage companies cannot. A sudden change in employment law, payment regulation, data policy or foreign ownership can consume months of management time and scarce cash.

For founders, uncertainty can be more damaging than strict regulation.

Stability does not mean no change

Governments must update policy.

Artificial intelligence, digital finance, platform work and cybersecurity all require new rules. A government that refuses to adapt can become as unattractive as one that changes direction too often.

The real difference is between structured reform and improvisation.

Credible governments explain why a rule is changing. They consult companies, coordinate between agencies and provide transition periods. Businesses may dislike the new rules, but they understand what is expected.

Less predictable systems announce policies abruptly, issue incomplete guidance or allow different agencies to interpret the same rule in different ways.

Investors can adapt to change. They struggle with confusion.

Also Read: Asian investors aren’t choosing between crypto and TradFi anymore

Vietnam: Consistency of direction

Vietnam is not Southeast Asia’s easiest market.

Businesses still report licensing delays, infrastructure constraints and differences between national and provincial implementation.

But Vietnam has maintained a clear economic direction for decades.

Successive governments have supported export-oriented manufacturing, trade integration and foreign investment. The details have evolved, but the broader strategy has remained recognisable.

That consistency has helped Vietnam build deep manufacturing supply chains.

Electronics companies attract component suppliers. Suppliers create demand for logistics, industrial software and professional services. Workers gain technical experience. Some later become founders or investors.

Vietnam is now trying to move into semiconductors, advanced electronics and higher-value manufacturing.

This shift will be difficult. Skills, energy supply and infrastructure remain constraints. But investors can see that the new strategy builds on the country’s existing industrial base.

Vietnam’s advantage is not perfect regulation. It is confidence in the long-term direction.

Malaysia: The challenge of implementation

Malaysia has strong infrastructure, experienced industrial clusters and an established role in electronics and semiconductors.

It also has a history of launching ambitious plans that can become harder to follow across political transitions and overlapping government agencies.

The country is now trying to build more durable industrial institutions.

The New Industrial Master Plan 2030 focuses on advanced manufacturing, semiconductors, technology and decarbonisation. The Johor-Singapore Special Economic Zone is another major test.

The zone aims to combine Singapore’s capital and connectivity with Johor’s lower costs, available land and workforce.

The economic logic is strong.

The challenge is execution.

Companies will judge the project by whether customs, immigration, licensing and investment approvals actually become simpler. They will also ask whether commitments survive changes in ministers and government priorities.

Malaysia does not lack strategies. Its competitive advantage will depend on turning those strategies into systems that businesses can trust.

Also Read: Reverse home bias: Why Southeast Asia’s digital investors may be diversifying in the wrong direction

Singapore: Credibility as infrastructure

Singapore offers the region’s clearest example of regulatory predictability.

Its rules are not always light. Financial services, employment, data protection and corporate governance are closely regulated.

Its advantage lies in the process.

Changes are usually announced clearly, accompanied by guidance and introduced through institutions with defined responsibilities.

This gives investors confidence that official decisions, contracts and regulations will retain their meaning.

Singapore’s model also has limits.

It is expensive. Land is scarce. Labour costs are high. The domestic market is small.

As a result, many companies place headquarters, intellectual property and financing functions in Singapore while locating manufacturing or operations elsewhere in Southeast Asia.

This shows the value of regulatory credibility. Even when physical activity is distributed across the region, ownership and strategic control often remain in the jurisdiction investors trust most.

Stability alone is not enough

Policy stability can also preserve bad systems.

A predictable but inefficient licensing process is still inefficient. Stable protectionism can still discourage investment. A long-standing subsidy may support weak companies rather than productive ones.

Consistency therefore needs to be combined with competence.

Governments must be able to update policies, enforce them fairly and coordinate across agencies.

Growth can also compensate for instability.

Investors may accept regulatory risk in markets with exceptional consumer growth, strategic resources or strong supply-chain advantages.

Also Read: India’s IPO boom is rewriting the exit playbook for global investors

But this often influences the type of capital that arrives.

Short-term investors may tolerate uncertainty. Factories, infrastructure projects and research centres cannot move easily once established. They require greater confidence in the future.

Policy predictability matters most when a country wants long-term capital that trains workers, develops suppliers and becomes embedded in the local economy.

What this means for startup ecosystems

Startup policy is often built around visible programmes.

Governments announce accelerators, matching funds, conferences, tax incentives and startup visas. These initiatives can help, but they do not create an ecosystem on their own.

Founders also need reliable company law, sensible tax treatment of employee shares, predictable visa rules, workable bankruptcy procedures and clear data regulations.

When these systems are uncertain, founders adapt.

They incorporate holding companies abroad. They keep intellectual property in Singapore. They hire through foreign entities. They raise capital in another jurisdiction.

The startup may continue operating locally, but ownership, financing and strategic control move elsewhere.

Countries then risk retaining low-value activity while losing the parts of the company that create the most wealth.

Credibility may be the cheapest incentive

Southeast Asian governments are competing with tax holidays, grants, industrial zones and infrastructure spending.

But incentives lose value when investors do not trust the policy framework around them.

A 10-year tax concession is less attractive if its interpretation may change after three years. A startup visa is less useful if approvals are inconsistent. A digital strategy means little if companies cannot determine which agency controls implementation.

Governments do not need to promise that rules will never change.

They need to show that change will be explained, coordinated and introduced through a process companies can understand.

That commitment requires administrative discipline more than public spending.

As Southeast Asia competes for factories, data centres, venture capital and technology companies, growth will remain the first attraction.

Predictability will increasingly decide where investors stay.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post Investors do not just fund startups. They fund predictability appeared first on e27.