Day: October 7, 2026

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.

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.

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.