Two weeks ago, I sat across from the new chairman of a cooperative in West Java. He had been appointed three months earlier under Indonesia’s Koperasi Merah Putih programme. He asked me, sincerely and without embarrassment, how he was supposed to know which risks his cooperative was actually carrying — and whether he had to file a report about it this quarter.

I have spent fifteen years inside Indonesian risk functions — banking, insurance, sharia microfinance — and I have heard versions of that question before. But never at this scale, and never with this little time to answer it.

That conversation is happening in 83,000 cooperatives across Indonesia right now. Each one has been required, under the Prabowo administration’s flagship Koperasi Merah Putih initiative, to implement formal risk management on day one — to identify the risks it is carrying, document the controls in place against them, monitor early warning signs, and record incidents as they happen. For most of these institutions, it is the first time anything that could be called governance has been written down.

The scale of the rollout is unlike anything ASEAN has attempted before.

The unprecedented scale

To put 83,000 cooperatives in context, it is roughly seventy times the number of commercial banks in Indonesia, and more institutions than there are public companies on the Indonesia Stock Exchange. The combined economic activity flowing through these cooperatives, even at modest per-unit volume, will touch tens of millions of households inside two years.

Indonesia has run financial inclusion experiments at scale before. Microfinance, sharia banking, branchless banking — each one produced lessons the region eventually absorbed. But none of them required formal enterprise risk frameworks on day one. The Koperasi Merah Putih programme is the first time a population-scale financial inclusion initiative has been launched with risk management embedded as a prerequisite, not as a maturity stage.

That decision is consequential. It is also extraordinarily ambitious.

Also Read: Business judgment on trial: Indonesia’s corruption courts are getting it backwards

What can go wrong

Three failure modes are predictable enough that they deserve to be named while there is still time to design around them.

Paper compliance. With deadlines this tight, the easiest response for cooperative leadership is to download a template, fill in the fields, file it, and move on. The documents exist on paper, but nobody on the ground is using them to make a decision. Within twelve months, they are stale, the staff have moved on, and the framework that was supposed to govern operations has become a binder in a drawer.

Supervisory dilution. Indonesia’s regulators — OJK, Kementerian Koperasi UKM — are themselves resource-constrained. Supervising 83,000 newly-launched institutions to a standard that takes years to build inside a commercial bank is, realistically, not going to happen at full depth. The risk is that the framework exists in policy but is enforced inconsistently, which is the worst of both worlds: cost without protection.

Loss-event blindspots. Cooperatives sit closer to their members than commercial banks do. They will be exposed to risks that traditional banking frameworks do not measure well — local social capital risk, agricultural cycle risk, informal credit chain contagion. A framework written in the language of banks will under-detect the things cooperatives are actually exposed to.

What needs to be in place

The next eighteen months will decide whether this becomes the largest financial inclusion success in Southeast Asia’s history or its most expensive policy lesson. Three things will determine which.

Training depth must outrun the deadline. Cooperative leaders need apprenticeship support, not certification cycles. Practical, hands-on coaching from people who have actually run risk frameworks inside real institutions — not slide decks — is what turns paper compliance into actual practice.

Also Read: Business judgment on trial: Indonesia’s corruption courts are getting it backwards

Tooling must be priced for the unit. The compliance software that costs ninety thousand dollars a year inside a large bank cannot be the model for a cooperative serving two thousand members. The tooling that works at this scale is cloud-native, modular, priced in tens of dollars per month, and operable by someone without a finance degree. The price point is, as it turns out, the easy part. The hard part is making the framework legible to someone who has never read a policy document before.

Reporting must aggregate upward. The supervisory burden cannot be solved by visit-each-cooperative auditing. It will only be solvable by data flowing up — from cooperative to district to regional to national — so regulators can spot anomalies at scale. That requires a deliberate reporting backbone, not an ad-hoc PDF submission system.

What ASEAN should learn from watching

The Philippines has cooperatives and rural banks facing similar inclusion-with-governance tensions. Malaysia and Thailand have parallel structures. Vietnam is beginning to formalise its financial cooperative sector. None of them has yet tried risk management at this scale on day one.

If the Indonesian experiment works, the model will be exported across the region within five years. The first institutions, supervisors, and operators to make it work at scale here will have an outsized influence on how the rest of ASEAN learns to do this.

If it fails, the lesson will be the more important one. Mandating governance at scale, without proportionate investment in capability and tooling, does not produce governance. It produces forms.

Either way, eighteen months from now we will know which of those outcomes Indonesia chose. The rest of ASEAN should be paying attention right now — because their version of this question is coming next.

—

The post The 83,000 experiment: Indonesia is running ASEAN’s largest test of risk management at scale appeared first on e27.

Manus’s reported buyback encapsulates a new reality: AI deals are no longer just commercial transactions but geopolitical chess moves with clear implications for Southeast Asia.

If Chinese investors repurchase Manus at the roughly US$2 billion price Meta paid, the outcome will reflect Beijing’s intent to retain control over advanced AI assets developed with Chinese capital, even when those assets are domiciled in neutral hubs such as Singapore.

That matters for Southeast Asia because the region is both a talent pool and a battleground for influence: Singapore remains an attractive incorporation point, but regulatory pressure from neighbouring powers can reshape ownership, governance and market access overnight.

The Manus story also exposes practical frictions for founders and VCs. Rapid revenue growth makes firms more strategically valuable, yet it also raises the stakes in national-security reviews. Western funds face difficult choices about exits and follow-on strategy; regional investors may see opportunities to repatriate value but must manage international credibility and market access.

For Southeast Asia’s startup ecosystem, the takeaway is twofold: structure deals with geopolitical scenarios in mind, and expect capital to flow along new corridors, including Hong Kong listings and China-centric joint ventures. Manus is not just one company’s reversal; it is a preview of how AI capital and control will be negotiated across Asia.

REGIONAL

Chinese investors move to unwind Manus’s Meta deal at US$2B: The attempted reversal reflects how US-China tech tensions are forcing Chinese-backed startups to choose sides and how deals once seen as validation are now becoming liabilities for founders navigating geopolitical crossfire.

BRI Ventures case: four executives jailed, no personal gain proven: The conviction despite no proven personal enrichment sets a troubling precedent for state-backed investors in Indonesia, where the line between bold investment decisions and criminal liability now appears dangerously unclear.

Carro acquires Australian firm CarPlace in cross-border push: The deal marks the Singapore-based used-car marketplace’s first move outside Southeast Asia, linking two fragmented automotive resale markets and raising questions about whether other SEA platforms will follow with similar outbound plays.

Vertex-backed ACRAB closes US$350M for agentic AI infra: One of the largest AI infrastructure deals in Southeast Asia this year, the raise positions ACRAB to compete directly with hyperscalers on compute supply for enterprise agentic workloads across the region.

Singapore AI startup K25 closes US$10M pre-Series A: With pre-A funding secured and Series A already underway, K25 AI is moving quickly, a sign that enterprise AI in Singapore is still attracting early-stage capital despite a broader global venture slowdown.

Respond.io raises fresh capital in latest funding round: The customer messaging platform will use the new capital to deepen its foothold among Southeast Asian enterprises, competing in a market where fragmented messaging channels remain a persistent pain point for regional businesses.

Golden Gate Ventures opens office in Uzbekistan: The Singapore VC’s first Central Asian outpost is a bet that the region’s emerging tech firms are ready to scale into Southeast Asia, and that capital flows between the two corridors can be meaningfully accelerated.

100×100 bets US$100M on 50 climate startups in SEA, India: The fund is structured as a startup factory, backing 50 companies at early stage across two of Asia’s most climate-vulnerable markets, where institutional climate capital has historically been thin.

Singapore leads APAC in AI agent rollouts and rollbacks: The findings suggest deployment speed is outstripping governance readiness, a pattern that could expose enterprises to operational and compliance risk as agentic AI moves deeper into business-critical functions.

MAS chief flags AI risk even as Singapore’s economy holds firm: The central bank governor cautioned that AI-related financial risks , including model opacity and systemic concentration, could undermine stability even as Singapore’s near-term economic outlook remains relatively resilient.

Singapore urges ASEAN to pursue AI without ceding data sovereignty: At a regional forum, Singapore pushed fellow ASEAN members to adopt AI collectively while guarding against the data dependency and sovereignty risks that come with over-reliance on a small number of foreign AI infrastructure providers.

Singapore workers adopt AI faster than their bosses: The gap threatens to create a two-tier workforce where employees build AI fluency that management cannot evaluate, undermining the strategic oversight needed to deploy these tools responsibly at scale.

Singapore launches US$29M scheme to fund digital media content: The programme targets media professionals transitioning into digital content creation, reflecting the government’s broader push to future-proof creative industries as traditional media revenue models continue to erode.

Vietnam and Singapore lead Southeast Asia in construction tech: Both markets are deploying construction technology at a pace that outstrips the rest of Southeast Asia, driven by large-scale infrastructure pipelines and a growing base of proptech and built-environment startups.

Indonesia runs ASEAN’s largest risk management experiment: The 83,000-participant pilot has direct implications for insurtech and financial inclusion models across ASEAN, where underinsurance and informal labour markets make population-scale risk programmes both necessary and enormously difficult to execute.

Vietnam’s growth hinges on structural dependence on private capital: Unlike peers that treat foreign investment as supplementary, Vietnam has built its economic growth model around it, a distinction that raises the stakes considerably if investor sentiment shifts or geopolitical conditions tighten.

---

INTERVIEWS & FEATURES

Tin Men Capital on backing unglamorous but durable SEA bets: The firm argues that Southeast Asia’s most defensible businesses are not the most visible ones; its portfolio strategy deliberately targets sectors with high switching costs and low VC competition, where patient capital can compound quietly.

How Marsham Edge is rethinking AI anomaly detection: The startup’s approach treats anomaly detection as a continuous learning problem rather than a rules-based one, targeting financial and logistics clients in Southeast Asia where irregular patterns often go undetected until material damage has occurred.

Vietnam’s biggest PE deal of 2025 was a food company: The outcome challenges the tech-first narrative that dominates SEA investor conversations; patient capital is quietly finding its best returns in essential consumer goods, not software.

---

INTERNATIONAL

OpenAI recruits senior talent ahead of anticipated IPO: The hires span finance, legal, and communications — roles that reflect a company shifting its centre of gravity from research output to investor relations, regulatory compliance, and public market readiness.

PayPal Ventures shuts down amid broader company restructure: The closure ends a funding channel that had backed fintech startups across Asia and Latin America, reflecting a wider retreat by corporates from venture activity as balance sheet discipline takes precedence over strategic portfolio plays.

Telegram ban in India drives users to VPNs and rival apps: The crackdown is already reshaping messaging app dynamics across South Asia, with spillover effects likely in Southeast Asia where Telegram remains a primary channel for communities, traders, and political organising.

Waymo recalls 4,000 robotaxis over highway construction flaw: The software error caused vehicles to navigate into active construction zones, an edge case failure that regulators in multiple markets are likely to cite as evidence that autonomous vehicle certification standards need tightening.

AI pressures could force Apple to raise iPhone prices: Rivals are shipping more capable on-device AI while Apple plays catch-up, and the cost of closing that gap may be passed directly to consumers through higher handset prices, a significant risk in price-sensitive Southeast Asian markets.

---

CYBERSECURITY

Cybercriminals breach tens of thousands of Fortinet firewalls: The alleged intrusion targeted enterprise-grade perimeter security deployed by major global corporations, exposing a systemic vulnerability in firewall infrastructure that many Southeast Asian enterprises rely on as their primary network defence.

---

SEMICONDUCTOR

Renesas acquires Pictorus to accelerate embedded software development: The deal plugs a browser-based behavioural modelling tool into Renesas 365, enabling engineers to generate Rust, C/C++, and Python-compatible embedded code from block diagrams, strengthening the platform’s pitch to automotive and robotics developers.

ASML denies EUV breach after US raises China export control fears: Commerce Secretary Howard Lutnick confronted ASML leadership over whether its most advanced lithography equipment had circumvented export controls, a more serious allegation than its routine China business, which already bars EUV shipments under Dutch rules.

Apple-Intel chip deal is thinner than Trump’s announcement suggests: Reported details point only to a preliminary manufacturing arrangement, not a design partnership, but it still hands Intel’s floundering foundry business a credibility boost, backed by the US government’s 9.9% stake and US$8.5B in prior grants to the chipmaker.

Amazon moves to sell its own AI chips to challenge Nvidia: Selling Trainium and Inferentia externally would let Amazon monetise its chip investment beyond its own cloud, directly threatening Nvidia’s grip on the accelerator market at a moment when enterprises are actively seeking supply alternatives.

---

AI

AI did not kill creativity; it just raised the bar: The argument is that generative tools have commoditised baseline creative output, making distinctly human judgement, taste, and originality more valuable, not less, for individuals and teams willing to develop them.

Only 16% of Americans expect AI to benefit society: The data points to a widening trust deficit that tech companies and policymakers have yet to address meaningfully, a sentiment gap with real consequences for AI product adoption, regulation, and public legitimacy.

---

THOUGHT LEADERSHIP

The higher you rise, the less you hear, and the more it costs: Organisations systematically filter out uncomfortable truths as they move up the chain, leaving senior leaders making high-stakes decisions on increasingly sanitised information, a structural failure, not a personal one.

Bitcoin’s 81% gold correlation signals a new macro identity: The data suggests Bitcoin is being absorbed into institutional portfolios as a macro hedge rather than a speculative asset, a shift with significant implications for how crypto fits into diversified portfolio construction.

When gold, stocks, and crypto fall together, nothing hedges: Classic diversification theory assumes low correlation between asset classes, but simultaneous declines across all three expose a structural flaw in modern portfolio construction that neither retail nor institutional investors have adequately prepared for.

Why tech giants are crashing as Bitcoin surges past US$67K: The divergence challenges a long-held assumption that risk assets move in tandem, suggesting capital is rotating out of equities and into crypto as a distinct macro hedge, not merely following the same sentiment cycle.

Why Pop Mart’s Labubu loyalty is not the same as brand loyalty: Consumers are devoted to the character, not the company — a vulnerability that leaves Pop Mart exposed if Labubu’s cultural moment fades, with no deeper brand architecture to fall back on.

The post Ecosystem Roundup: Manus buyback is a geopolitical wake‑up call for SEA’s AI ecosystem appeared first on e27.

Singapore-based enterprise AI company fileAI has received a strategic investment from JRE VENTURES, the corporate venture capital arm of Japan’s JR East Group, marking a significant step in the company’s expansion into the Japanese market.

The investment comes alongside a deployment partnership in which fileAI’s governed AI platform will be rolled out across JR East Group companies. The JR East Group operates one of Japan’s largest and most complex rail and transportation networks, making the partnership a high-profile test of the platform’s enterprise capabilities.

fileAI’s platform deploys proprietary AI agents to convert legacy contracts and documents into structured, searchable knowledge assets. The tech spans document digitisation, AI-powered data extraction, centralised repositories, and contract intelligence — addressing a longstanding challenge for large organisations that hold vast archives of paper-based and legacy digital records.

Under the partnership, fileAI and JRE VENTURES aim to digitise historical contracts and documents, automatically extract key terms, clauses, obligations, and metadata, consolidate that knowledge into a centralised intelligent repository, and generate contract analytics including risk insights and renewal forecasting.

Also Read: The 83,000 experiment: Indonesia is running ASEAN’s largest test of risk management at scale

The long-term ambition is to establish what fileAI describes as a “living contract intelligence layer” across organisations, reducing manual document handling and enabling more informed operational decision-making.

Christian Schneider, chief executive of fileAI, described Japan as a pivotal market for the company. “Their appetite for innovation, combined with the scale of their operations, makes this the perfect proving ground for what AI agents can do for enterprises,” Schneider said in a statement. fileAI is currently building a dedicated local team in Tokyo, with hiring under way across sales, engineering, and customer success.

Junichi Eto, managing director of JRE VENTURES, said the partnership would help validate practical use cases for enterprise AI in Japan and the broader Asia-Pacific region. “fileAI’s approach to AI-driven file processing represents a meaningful advancement in how enterprise data can be structured and utilised,” he said.

The investment was facilitated through 1982 Ventures, a Singapore-based fund manager focused on enterprise AI, fintech, and private markets across Asia. The firm positioned itself as a bridge between high-growth Asian technology companies and Japanese corporate investors.

Also Read: Chinese backers move to buy Manus from Meta in potential US$2B reversal

Herston Powers, founding managing partner at 1982 Ventures, said the JR East Group deployment offered strong market validation. “Every large enterprise sits on a goldmine of trapped data. Seeing them deploy inside the JR East Group — a massive, complex environment — is the best kind of validation,” he said.

fileAI did not disclose the financial terms of the investment.

—

Image Credit: fileAI

The post fileAI secures strategic investment from JR East Group’s venture arm to expand in Japan appeared first on e27.

K25.ai, an APAC-focused startup attempting to fuse live streaming, creator monetisation, and prediction markets, has closed a US$10 million pre‑Series A round at a US$100 million valuation.

With this deal, Nasdaq-listed NewGenIVF Group completed a US$4 million tranche that brings its aggregate commitment in the AI firm to US$10 million. Once closing conditions are satisfied, NewGen’s ownership in K25.ai is expected to rise to roughly 10 per cent.

Also Read: K25.ai bags strategic funding from Nasdaq-listed NIVF at US$100M pre-A valuation

K25.ai has already begun raising a Series A to fund commercialisation and regional expansion.

A product for creator-first Asia

K25.ai bills itself as a fusion of Twitch, prediction markets such as Polymarket, and generative-AI assistants. This way, it provides an experience that lets creators host livestreams while audiences “watch-to-predict” outcomes of sports, e-sports and entertainment events. The startup says AI will help create and resolve markets, while enabling real-time community participation.

For a region where livestream commerce, creator monetisation, and e-sports are booming, the proposition is timely. Southeast Asia’s internet economy continues to expand, fuelled by mobile-first consumption and rising creator activity across platforms, such as TikTok and YouTube. Localised content, live engagement features and novel monetisation mechanisms matter more here than in many mature markets, a dynamic that plays directly into K25.ai’s stated strengths.

Asia Pacific is both culturally receptive to live, communal viewing experiences, and regulatory-complex when it comes to information markets. If they can thread the needle between product-market fit and compliance, there is a clear path to scale.

Strategic stake and regional agency rights

NewGen’s additional investment follows a US$2 million commitment in May and a further US$4 million announced on June 4, completing its headline US$10 million backing. Beyond a financial stake, NewGen has secured exclusive Asia Pacific agency rights with K25.ai, a commercial arrangement that grants it distribution and partnership opportunities across permitted markets.

That dual arrangement of equity plus agency rights is notable. It gives NewGen both an economic upside and a route to monetise the product regionally through local partnerships, distribution deals, and go-to-market activities.

Also Read: Streaming the dream: How live streaming technology can increase access to brands

For K25.ai, partnering with an investor that already has regional ties can accelerate market entry, particularly in jurisdictions where navigating regulatory regimes and building creator ecosystems are resource‑intensive.

NewGen’s chairman and CEO Alfred Siu framed the move as strategic: the firm is seeking exposure to “a differentiated platform operating at the convergence of artificial intelligence, livestreaming and prediction-market infrastructure.”

K25.ai’s CEO Andy Cheung said the pre‑A close is “strong strategic validation” and that the Series A will accelerate product launch and regulatory licensing in selected Asian markets.

Regulatory obstacles and the compliance imperative

Prediction markets and wagering-adjacent products face a patchwork of legal regimes across Southeast Asia. Countries such as Singapore and Malaysia impose strict rules on gambling and speculative betting, while others adopt more permissive frameworks for information markets or skill-based prediction activities. K25.ai says it will not operate in jurisdictions where such activities are restricted or prohibited, and it is pursuing applicable regulatory licensing in selected markets.

That cautious stance is necessary: several startups in the prediction and betting space have run into regulatory pushback when launching without sufficient local licences or when their product crossed into gambling territory. For K25.ai, the technical promise of AI-assisted market resolution needs to be matched by clear legal boundaries and robust age‑gating, geofencing and compliance tooling, especially if creators in multiple countries can host events that attract cross-border audiences.

The Series A will likely be as much about compliance and localisation as it is about marketing and creator acquisition. Investors in the region will want to see concrete plans for regulatory approvals, partnerships with licensed operators where required, and product controls that demarcate entertainment from gambling.

Monetisation and creator economics

K25.ai’s model ties revenue potential to creator engagement and the liquidity of prediction markets: higher viewership and more active markets can translate into fees, sponsorship deals and potentially secondary markets for data and signals. Southeast Asia’s creators are expert at converting live engagement into commercial outcomes — think live commerce on Shopee or TikTok — but prediction-based monetisation is newer.

Monetisation hurdles include user education, trust in market settlement and the need to seed liquidity so markets feel meaningful. AI can help standardise market creation and speed up resolution, but user-facing clarity and transparent settlement mechanics will be crucial to adoption. Given the region’s appetite for esports and fantasy sports, these categories could be early wins if regulatory fit is achieved.

Why the Southeast Asian angle matters

Southeast Asia represents a large and young digital-native audience, a flourishing creator economy and high mobile engagement, all tailwinds for a live-streaming, prediction-driven product. Local languages, cultural nuances in content, and varying regulatory regimes mean success will depend on granular market-by-market strategies rather than a one-size-fits-all regional roll‑out.

NewGen’s agency rights across Asia-Pacific suggest K25.ai will lean on partners with established regional networks. That could speed creator recruitment, secure local licences faster and build a distribution play that a pure-play Silicon Valley investor might struggle to execute.

What to watch next

In the coming months, the market will be watching for K25.ai’s product launch cadence, the specifics of its Series A valuation and investor mix, and the company’s progress on regulatory approvals in key Southeast Asian markets. Execution on creator partnerships, user safety and market liquidity will determine whether the startup can turn its concept into a viable, scalable business.

Also Read: Live-streaming done right: How brands can turn viewers into loyal customers

For incumbents and investors, the combination of AI, live video and prediction mechanics is a fresh experiment. If K25.ai and NewGen can navigate regulatory complexity and prove a compelling creator revenue path in Southeast Asia, they could lay the groundwork for a new digital entertainment category across the region.

The post NewGen doubles down on K25.ai as Asia-focused AI livestreaming platform eyes commercial launch appeared first on e27.

The AI hype cycle loves a clean split: innovators and laggards. Southeast Asia’s story is messier and more interesting.

A study titled “AI in Southeast Asia: An era of opportunity” by McKinsey and the Singapore Economic Development Board surveyed 330 respondents across six economies — Indonesia, Malaysia, the Philippines, Singapore, Thailand, and Vietnam — and found the region is nudging ahead of the global average in moving beyond experimentation.

That is the good news.

Also Read: Southeast Asia’s AI boom is built on steel, not startups

The bad news is that the region’s economic backbone—micro, small, and medium-sized enterprises (MSMEs)—could be priced out of the next productivity leap unless AI becomes cheaper, simpler, and more local.

The adoption numbers: slightly ahead of the world, behind the US

According to the report’s regional breakdown, AI adoption in Southeast Asia is now heavily weighted toward scaling rather than dabbling. The data shows:

• 8 per cent fully scaled
• 38 per cent scaling
• 35 per cent piloting
• 19 per cent experimenting
• effectively negligible “no use at all” in the dataset

In other words, 46 per cent are beyond pilots (fully scaled + scaling). That edges the global composite in the report, and signals that “AI in enterprise” is no longer exotic in the region’s more digitally advanced markets.

Yet the US still leads, with higher fully scaled and scaling shares. Southeast Asia is not winning on maturity; it is winning on momentum.

Size matters and it’s not subtle

The report slices adoption by company revenue. The pattern is predictable, but the gap is still meaningful:

• Large firms (annual revenue more than US$250 million): 56 per cent
• scaling or fully scaled
• Medium firms (US$100 million–US$249 million): 47 per cent scaling or fully scaled
• Small firms (less than US$100 million): 42 per cent scaling or fully scaled

That is the real divide: not country, not sector, but organisational capacity. Larger enterprises have deeper data pools, more stable infrastructure, and budgets that can absorb mistakes. Smaller businesses have less room to “learn by doing” when the learning curve costs money.

MSMEs are in the region. AI pricing could decide who wins

Southeast Asia has 70 million MSMEs, the report says, representing about 97 per cent of the workforce and a large share of GDP. That makes AI adoption for small firms less a technology question than an economic one.

If AI tools remain priced and packaged for enterprise procurement teams, the region gets an ugly outcome: big firms compound their productivity advantages while small firms fall further behind, even if the technology itself is “available”.

Also Read: Rethinking AI adoption: Why Southeast Asia’s businesses must transform to thrive

The report calls out what MSMEs need from providers:

• low-cost entry options
• local currency pricing (or at least predictable usage-based pricing)
• bundled packages (collaboration tools + data + model access + onboarding)
• guided adoption to reduce complexity

That is basically a demand for AI as a utility, not AI as a bespoke transformation programme.

The sector leaders are what you’d expect and that’s the point

AI maturity varies by industry. The report highlights technology, media, telecommunications, and advanced industries as the leaders, with around six in ten firms scaling or fully scaled. Energy and materials also show substantial progress, with around half of them scaling.

In contrast, the public sector, healthcare, travel, and infrastructure remain earlier-stage, with over six in ten still piloting or experimenting. This is not because those sectors lack use cases. It’s because they have nastier data environments, heavier regulation, and higher consequences when models hallucinate or leak.

Real adoption is changing job expectations — not just dashboards

The report includes a candid Grab quote that reveals what “AI adoption” actually looks like inside a scaled platform.

Grab’s group head of data and analytics, Nikhil Dwarakanath, says: “We have several implementations that are running at scale, such as our merchant AI assistant, now rolled out to over 1.2 million merchants…”

He adds: “AI is helping to improve top-line growth. For example, merchants using the merchant assistant have seen their business grow by about 10 per cent.”

That is a direct claim of revenue impact from a scaled AI product. It also hints at a regional opportunity: platforms that serve MSMEs can act as AI distribution rails, delivering AI benefits to small businesses that would never build these systems on their own.

People are unusually optimistic about AI here. That’s an advantage

One of the report’s more striking societal stats: 70 per cent of the population in Southeast Asia regard AI as a societal benefit, compared with 44 per cent in Japan and 42 per cent in the US (as cited in the report).

This matters because adoption is not just about budgets and infrastructure; it is about trust and willingness. A region that is culturally open to AI products may see faster consumer uptake and less friction in deploying AI-enabled services—especially in mobile-first markets.

The real bottleneck is not curiosity — it’s operational discipline

Southeast Asia’s AI adoption is no longer stuck at “pilot theatre”. But scaling beyond pilots is not the same as scaling impact. The next stage will be determined by whether companies can:

Also Read: From hesitation to action: How SMEs in Southeast Asia can start AI adoption

• integrate AI into messy legacy systems
• build or buy the right talent (especially MLOps and applied engineering)
• prove ROI beyond productivity anecdotes
• manage risk without paralysing deployment

The region’s momentum is real. But momentum alone does not produce winners. Pricing models, packaging, and platform distribution—especially for MSMEs—could decide whether Southeast Asia’s AI wave becomes broad-based growth or just another round of consolidation for the biggest players.

—

The image was created using AI.

The post Momentum without maturity: Southeast Asia’s AI reality appeared first on e27.