Prelude

The global economic landscape is currently navigating a period of profound structural realignment, characterised by the aggressive internationalisation of Chinese small and medium-sized enterprises (SMEs). This movement, colloquially termed “ChuHai” (出海) or “Going to Sea,” has transitioned from a tactical response to domestic market saturation into a foundational strategic imperative for the survival and long-term viability of the Chinese private sector.

As we approach the late 2020s, the ChuHai phenomenon is no longer merely about exporting low-cost commodities; it represents a sophisticated evolution toward brand excellence, localised operational capacity, and the deployment of advanced technological ecosystems across emerging and developed markets alike.

This report provides an exhaustive analysis of the drivers, magnitude, and regional complexities of this trend, while identifying high-potential market opportunities for new ventures seeking to support this massive wave of globalisation.

The structural drivers of overseas expansion: The involution crisis

At the core of the current surge in Chinese SME internationalisation is the phenomenon of “involution” (内卷, nei juan). This term describes a state of hyper-competition where excessive effort and resources are expended for diminishing returns, often leading to a race-to-the-bottom in pricing and profit margins. The structural roots of this crisis are multi-dimensional, involving fiscal pressures, demographic shifts, and the collapse of the traditional growth engines that fueled China’s rise over the previous three decades.

The exhaustion of the real estate sector, which historically accounted for 20 per cent to 30 per cent of China’s GDP and 27 per cent of all bank loans, has created a massive vacuum in the domestic economy. The bursting of property bubbles has not only eroded household wealth and suppressed consumer demand but has also stripped local governments of their primary source of fiscal revenue: land financing.

Also Read: Trust takes years to build but one flawed system can damage a micro business overnight

By late 2025, local government debt had escalated to an estimated US$18.9 trillion, forcing these entities to pivot toward manufacturing and high-tech sectors as alternative drivers of GDP growth. This pivot has resulted in a deluge of government subsidies, tax incentives, and low-interest loans directed toward state-favoured industries, including electric vehicles (EVs), solar equipment, and semiconductors.

The unintended consequence of this state-led investment has been a chronic oversupply and massive overcapacity. When domestic demand failed to keep pace with the state-subsidised production surge, firms were forced into brutal price wars to survive. In the EV sector, for example, dominant players have used financial leverage to pursue predatory pricing strategies intended to eliminate smaller competitors.

By late 2025, industrial profits in several manufacturing segments saw year-on-year declines as sharp as 13.1 per cent, effectively erasing previous growth and creating a “growth without profits” trap. For many SMEs, the domestic market has become a zero-sum game, making international expansion the only viable pathway for maintaining operational solvency and achieving sustainable margins.

Market size and profiling: The 2026-2028 surge

The magnitude of the Chinese SME overseas surge is reflected in the record-breaking metrics of outward direct investment (ODI) and the volume of private enterprises engaging in global trade. By the end of 2025, the number of private Chinese enterprises with actual import and export activity reached approximately 613,000, accounting for the vast majority of the country’s 700,000 active trade entities.

Also Read: The divided AI race nobody wins: How businesses can navigate the US-China tech divide

For the 2026-2028 cycle, the “ChuHai” market is expected to expand by approximately 175,000 SMEs annually. This “New Wave” is characterised by a transition from “Made in China” (volume export) to “Operated by China” (localised presence).

SME internationalisation profile (2026-2028)

Strategic expansion priorities

Chinese SMEs are no longer pursuing simple volume; their expansion is now “capability-led,” focusing on the following strategic pillars:

• Market expansion: Escaping domestic deflation and price wars to capture margins that are often double what is achievable domestically.
• Technology licensing and IP: Shifting toward licensing proprietary technology to local partners to overcome regulatory barriers and secure data exclusivity in sensitive sectors like biomedicine and AI.
• Global R&D and talent: Establishing overseas innovation centres to access bilingual leadership and local technical talent, bridging the cultural gap between HQ and the market.
• Manufacturing outsourcing and nearshoring: Relocating production capacity to regions like Mexico (nearshoring) or Vietnam to bypass US and EU tariffs and shorten delivery cycles from weeks to days.

The role of the accelerator state: Policy support and institutional frameworks

The internationalisation of Chinese SMEs is a core component of the national industrial strategy. The government has evolved into an “accelerator state,” moving toward a multi-layered system designed to fast-track the growth of high-tech SMEs in strategic sectors.

The Little Giants initiative

The “Little Giants” program focuses on “specialised, refined, special, and new” SMEs within key industrial chains. For the 2024-2026 period, the program prioritises the “six foundations”: core basic parts, core basic components, key software, advanced basic processes, key basic materials, and industrial technology foundations.

Capital support is significant, with guidelines aiming to inject up to CNY 6 million (approximately US$830,000) per firm over a three-year cycle. By late 2025, the program had cultivated over 13,000 national-level Little Giants, with cities like Shenzhen housing over 1,000 such firms.

Also Read: The scale layer nobody budgeted for: How AI agents unlock growth for Asian businesses

The 15th five-year plan and the 2030 horizon

The strategic roadmap for the next phase (2026-2030) outlines a shift from growth driven by scale to growth driven by quality. Key objectives include:

• Technological self-reliance: Accelerating breakthroughs in brain-computer interfaces, quantum technology, and semiconductor supply chains.
• Digital economy expansion: Increasing the share of core digital industries to 12.5 per cent of overall GDP.
• Support for global scale: Explicitly encouraging internet platforms, AI companies, and professional services to expand and form partnerships overseas.

Geographic realignment: Emerging corridors and the global South

As regulatory scrutiny intensifies in the US and EU, Chinese SMEs are diversifying toward regions with lower regulatory friction.

ASEAN: The hub-and-spoke model

ASEAN is the critical region for restructuring Chinese supply chains, with FDI inflows reaching US$226 billion in 2024. For Chinese SMEs, ASEAN offers a mobile-first consumer base aligned with Chinese digital strengths. Vietnam, Malaysia, Indonesia, and Thailand have become central hubs for manufacturing and customer service.

The Middle East: The Gulf blue ocean

The Middle East—particularly the GCC states—is a priority destination for Chinese capital. In 2024, the region received US$39 billion in BRI investments, a 102 per cent increase YoY. Saudi Arabia alone drew US$19 billion. Chinese firms view the Gulf as a “blue ocean” due to high policy flexibility and security.

Latin America: Mexico and the nearshoring shift

In Latin America, the focus is shifting toward Mexico as a gateway to the North American market. Chinese brands now account for 57 per cent of cars imported into Mexico as of early 2025. The “Trump Corollary” to the Monroe Doctrine creates headwinds, but the USMCA framework provides a significant duty-free advantage for Chinese SMEs that can successfully localise manufacturing in the region.

Also Read: AI agents and the new rules of business execution

The technological architecture of ChuHai: AI and agentic trade

The year 2025 has been identified as China’s “AI Agent Year,” marking the deployment of autonomous systems to manage global operations.

Agentic workflows in cross-border operations

Next-generation AI agents are being integrated into platforms like WeChat (OpenClaw framework) to solve operational challenges. SMEs use “AI Agent Orchestration” to automate end-to-end marketing, content generation, and performance loops. In logistics, “Control Tower Agents” optimise delivery routes and reorder workflows, reducing routine task handling by 60-80 per cent.

The service ecosystem opportunity: Identifying business niches

The Chinese SME expansion has outpaced its supporting service ecosystem, creating massive gaps in talent, compliance, and localisation.

The talent gap: A staggering bottleneck

The number one bottleneck for Chinese companies going global is talent acquisition. In 2025, there was a talent gap of 4 million people in the cross-border e-commerce sector alone.

• Startup opportunity: AI-powered executive search for “bridge leaders” and Employer of Record (EOR) services to manage regional talent networks.

Compliance and regulatory readiness (C-as-a-service)

Compliance requirements for outbound firms surged by 250 per cent in 2025.

• Startup opportunity: Global payroll, HR compliance SaaS, and Data Sovereignty solutions to manage the web of local tax and privacy laws.

Localisation and ecosystem integration

Success is tied to the ability to “go in”—truly entering the local culture—rather than just “going out”.

• Startup opportunity: Cross-border traffic marketing, AI-native SEO (GEO), and market entry incubators for the Global South.

Also Read: The scale layer nobody budgeted for: How AI agents unlock growth for Asian businesses

Strategic conclusions and recommendations

The expansion of Chinese SMEs overseas is a structural trend that will define global commerce through 2030. Driven by the exhaustion of domestic profits and supported by a multi-billion-dollar state accelerator, this wave is moving toward higher-value sectors and deeper regional integration.

For new ventures, the most promising path is to become a “strategic enabler” for this outbound surge. The transition from “Made in China” to “Brands from China” represents the next great shift in the global economy. Those who can provide the cultural, regulatory, and technological bridges will be positioned at the heart of the world’s most dynamic trade corridor. The path forward lies in combining AI-native “intelligence” with the “empathy” required for deep cultural localisation.

—

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 post ChuHai: The business opportunity nobody in Southeast Asia is talking about appeared first on e27.

This is about freight. Specifically, why freight, logistics, and supply chain, an industry so gloriously unglamorous that people fall asleep mid-sentence just describing it, is turning into one of the most interesting places to build a company in Asia right now.

I know. Bear with me.

Few people want to work in logistics — that’s the point

When I tell people I run a logistics company, one of two things happens. Either their eyes glaze over, or they say something polite and immediately change the subject.

Logistics has an image problem. It’s not the industry you dream about at university. It doesn’t attract the same talent pipelines, the same VC attention, or the same media coverage as the sexier corners of tech. Nobody is writing breathless Substack posts about customs clearance and freight.

And yet, quietly, something is happening.

The global 4PL market is projected to grow at 8.1 per cent CAGR from 2025 to 2032, according to research commissioned by Wayfindr citing Market. In e-commerce specifically, that growth rate accelerates to 12 per cent CAGR over the same period, driven by surging cross-border trade, supply chain complexity, and the relentless expansion of direct-to-consumer brands into new markets.

E-commerce is eating retail across every market. Manufacturing is rapidly diversifying across Southeast Asia as brands shift supply chains out of China. And the technology layer that connects all of this, the visibility, the coordination, the data, has barely been built.

Also Read: Cybersecurity strategies for startups on a budget

The problem is not shipping, it is complexity

Here is what most people get wrong about logistics as a startup opportunity. They think the opportunity is in moving things faster. A better courier. A smarter warehouse. A cheaper freight rate.

That is not the problem.

The real problem is that scaling an e-commerce brand across multiple countries, with manufacturing in Vietnam or China, selling into the US, UK, and Europe simultaneously, involves dozens of moving parts, dozens of providers, and nobody whose job it is to be accountable for all of it at once.

A brand owner running a US$20 million e-commerce business should be thinking about product, marketing, and growth. Instead, they spend half their week chasing updates from a freight forwarder here, arguing with a warehouse over there about a stock discrepancy, and trying to figure out why their landed costs keep changing.

That is not a shipping problem. That is an orchestration problem. And orchestration is exactly where tech has enormous room to run.

Vietnam is not a trend — it is a structural shift

I spend a lot of time in Southeast Asia. A big segment of our team operates out of Vietnam, and what we see on the ground is not a passing wave. It is a genuine realignment of global manufacturing.

Brands that were 100 per cent China-reliant five years ago are now actively splitting production. Vietnam, Indonesia, Taiwan, and Thailand are absorbing that shift under the China+ strategy. And with it comes a whole new layer of complexity: new suppliers, new compliance requirements, new last-mile challenges, and new currency risk.

Also Read: AI is removing the co-founder bottleneck for early-stage startups

The market data reflects this reality. APAC is the fastest-growing region for 4PL adoption globally, projected at 10.8 per cent CAGR through 2032, growing from a US$19.7 billion market today to US$44.7 billion by 2032. That growth is being driven directly by the manufacturing shift, the rise of cross-border e-commerce, and the urgent need for smarter supply chain infrastructure across the region.

For the brands navigating this, the operational burden has never been higher. For the companies building technology and services to help them do it, the opportunity has never been larger.

The startup opportunity in Southeast Asia’s logistics sector is not just local. It is the infrastructure layer for global commerce.

The boring industries have the best defensibility

Here is something I learned coming from the oil and gas world before logistics: the industries that look boring from the outside are often the ones with the deepest moats.

An operator who genuinely understands how freight consolidation works out of Guangzhou, how Vietnamese customs changes every year, and how to design a landed-cost model that holds across six destination markets, that knowledge does not transfer easily. The complexity is the barrier. And the complexity, at the moment, is only increasing.

Tariff changes. Carbon reporting requirements. Cross-border regulatory divergence. Every one of these adds another layer that brands need help navigating, and another reason to build toward a model where one intelligent, tech-enabled partner is accountable for all of it.

This is what the fourth-party logistics model, 4PL, exists to do. And it is a model that is still, genuinely, in its infancy in Asia.

What is a 4PL?

A fourth-party logistics provider, or 4PL, is an independent, non-asset-owning partner that designs, manages, and optimises your entire supply chain, coordinating freight forwarders, warehouses, carriers, and technology under one roof. Think of it less like a supplier and more like a control tower: one point of accountability for everything that moves.

Also Read: Startups driving AI automation, fintech, and accessibility gather at Echelon Singapore 2026

The unsexy bet is often the right one

I started building in logistics because I saw a problem I could not stop thinking about. Not because it was fashionable. Not because investors were excited. Frankly, most of them were not.

We bootstrapped to eight figures without a cent of external funding, which I mention not to brag but to make a point: the fundamentals of the problem were strong enough that we did not need anyone to believe in the vision before the market proved it. The demand was real. The inefficiency was real. The gap was real.

The next decade of e-commerce growth in Asia is going to be built on infrastructure. Not just digital infrastructure, but the physical and operational infrastructure that moves real products from real factories to real customers. The companies that build the intelligence layer on top of that, the platforms, the visibility tools, the coordination systems, those are the companies that will be quietly, unglamorously, extraordinarily valuable.

So yes. Freight. Supply chain. Logistics.

I promise it’s more interesting than it sounds.

—

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 post Why the most boring industry in the world is quietly becoming a startup goldmine appeared first on e27.

I know what you’re thinking (pun intended). But this is not a sensational fantasy about mind reading. It is an extrapolation from real advances in AI, brain-computer interfaces (BCIs), and neuroscience. As systems for decoding neural signals and translating thought-related activity into digital output continue to improve, the question is no longer just what they can do on their own, but whether they could one day be networked, and to what extent. Although BCIs are no longer a novel concept, if such devices could communicate directly with one another, they might give rise to technological telepathy.

From private thought to usable input

Thoughts have historically remained private by default, becoming shareable only when forced through speech, writing, gesture, or code, each of which introduces delay, tension, and translation between intention and expression. Technological telepathy becomes consequential not because machines can literally read minds in a science-fiction sense, but because computing is progressively collapsing that gap, as BCIs, silent-speech systems, neural decoding models, and generative AI converge into a communications stack in which cognition itself becomes a usable input, suggesting that future networks may connect minds rather than merely devices or identities.

A longer lineage of brain-machine translation

This trajectory does not originate with ATR, AlterEgo, Neuralink, or the current “AI summer,” but extends through a longer history of rendering the brain legible to machines, beginning with Hans Berger’s EEG and its demonstration of non-invasive neural capture, continuing through José Delgado’s stimulation experiments, Alvin Lucier’s ”Music for Solo Performer” and its use of EEG for artistic control, Eberhard Fetz’s work on learned modulation of neural firing, and Jacques Vidal’s articulation of “brain-computer communication,” later made publicly tangible through BrainGate’s cursor control for paralysed patients and Kevin Warwick’s experiments in technological telepathy, all of which situate this internet of minds as a continuation rather than a rupture.

Institutional drivers and military interest

This history cuts across neuroscience, engineering, military funding, performance art, and public spectacle, with DARPA embedded as part of the field’s institutional structure, particularly through programs such as N3 that target high-performance neural interfaces for human-machine teaming, active cyber defence systems, and control of unmanned aerial vehicles, although no credible public evidence supports operational BCI-to-BCI communication in military or covert use.

Also Read: AI is irrevocably changing the tech landscape, and you are going to need a new map

Science fiction as conceptual groundwork

Science fiction anticipated the conceptual and social implications well before technical feasibility, as seen in Alfred Bester’s “The Demolished Man“ and its treatment of telepathy and social order, William Gibson’s “Neuromancer“ and “Johnny Mnemonic“ and their linking of neural systems to networked computation, Iain M. Banks’s neural lace, Ramez Naam’s infrastructural treatment of networked cognition, and Isaac Asimov’s “Foundation,“ alongside concepts such as hive mind theory and consciousness field theory, which framed expectations even as technological telepathy itself remains grounded in engineering rather than speculative human abilities.

What technological telepathy actually is

In practical terms, technological telepathy does not involve full extraction of continuous private thought, but instead consists of narrower capabilities such as decoding constrained visual categories from brain activity during sleep, reconstructing partial features of perceived or imagined images, or inferring silently articulated words from neuromuscular signals in the face and jaw, which are distinct but collectively indicate that communication technologies are moving upstream toward earlier stages of cognition.

In practical terms, current and near-term systems rely on specific device classes rather than abstract “mind reading,” including implanted electrode arrays that record neural firing directly from the cortex, non-invasive headsets based on EEG or functional imaging that capture aggregate brain activity, endovascular interfaces that access signals via blood vessels, and wearable EMG sensors placed on the face or jaw to detect subvocal speech, all of which produce partial, task-specific signals that must be decoded and interpreted through software, meaning that what is transmitted is not raw thought but a constrained, device-mediated representation of selected aspects of cognition.

The layered communication stack

A clearer understanding emerges when treated as a layered system, beginning with capture through electrodes, imaging systems, or wearable sensors that detect neural or neuromuscular activity, followed by decoding via machine-learning models that map signals to probable words, intentions, percepts, or categories, then mediation through software that filters noise, ranks interpretations, predicts continuations, corrects errors, and structures ambiguous biological signals into coherent output, and finally transmission to devices, other individuals, or networks, with AI functioning as the intermediary that translates between biological activity and digital meaning rather than enabling direct thought transfer.

From data to the “internet of minds”

Within this architecture, cognition-related data becomes processable in ways analogous to other network data while remaining qualitatively closer to thought itself, introducing the possibility that privacy breaches occur prior to completed expression, and implying that the dominant model will resemble cognitively mediated client-server communication rather than direct peer-to-peer telepathy.

Also Read: How fintech in Asia is enabling and making education affordable for everyone

Philosophical constraints on “pure thought”

Philosophical objections of Neuralink-like research, such as Slavoj Žižek’s 2020 talk at the University of Winnipeg, emphasise that conceptual thought does not exist independently of language, challenging the notion that “pure thought” can be transmitted without distortion and reframing linguistic imperfection as intrinsic to meaning rather than an obstacle to be eliminated.

Technical limits and partial decoding

Technical constraints remain substantial, particularly in the form of invasiveness, as high-performance BCIs often depend on implants placed in or near the brain, introducing surgical risk, long-term maintenance, and questions of removal and bodily autonomy, while less invasive approaches such as wearables or endovascular interfaces shift these tradeoffs without removing them, as illustrated by ATR and Yukiyasu Kamitani’s lab, whose dream decoding studies demonstrated category-level prediction of dream content under tightly constrained conditions rather than full reconstruction, thereby establishing partial permeability of internally generated experience without generalisability.

Silent speech and the boundary of intent

Alternative approaches, such as AlterEgo, focus on silent speech, relying on intentional subvocalisation and neuromuscular detection to create a clearer boundary between private thought and transmitted output, although current limitations in surface EMG prevent reliable decoding of inner-speech phonetic content, reinforcing that existing systems detect controlled signals rather than unrestricted cognition.

The fragility of intentionality boundaries

This boundary of intent, while conceptually important, remains technically and institutionally fragile, as systems may expand the definition of “intended” signals through software updates, model retraining, or error correction, and as movement toward decoding imagery, semantic content, and prelinguistic intention further complicates distinctions between thinking, rehearsing, and transmitting, with proposed safeguards such as learned cognitive protocols or mental “keys” likely to erode under pressures for efficiency and usability.

From research frontier to platform economy

The transition from research to commercialisation, evident in companies such as Neuralink, Synchron, and Paradromics, reframes neurotechnology as infrastructure rather than experiment, introducing business models that range from high-cost clinical hardware reimbursed through healthcare systems to platform-based software and institutional deployment in workplaces, defence settings, or insurer-managed care, and elevating neural data as a potentially valuable resource due to its proximity to intention before action, preference before declaration, and friction before expression.

Also Read: Beyond Silicon Valley dreams: Why Southeast Asia is rewriting the rules of tech for good

Expression as a hybrid artifact

In this context, expression becomes a joint product, as systems infer, correct, rank, and autocomplete outputs derived from incomplete signals, producing hybrid artefacts that blur the boundary between user intention and machine contribution, thereby shifting the problem from privacy alone to questions of authorship and authenticity, since output may already reflect negotiation between human cognition and computational prediction.

Why consent is insufficient

Consent, traditionally understood as sufficient ethical grounding, becomes inadequate when users do not stand outside the systems shaping their expression, requiring structural governance mechanisms such as purpose limitation, auditable mediation, rights of refusal, prohibitions on employer coercion, protected clinical boundaries, and legal remedies when machine output is misattributed to the user as fully self-authored.

Feedback loops and cognitive adaptation

Feedback dynamics further complicate the system, as users adapt to decoder behaviour, anticipate system completions, and potentially reshape their own cognitive patterns to improve legibility and control, creating a feedback loop in which thought is partially oriented toward machine interpretability and generating ambiguity in responsibility when outputs reflect inferred rather than explicitly intended meaning.

Assistive promise and differential cognitive citizenship

While assistive applications for paralysis, severe motor impairment, or speech loss remain one of the strongest justifications, variability in neural and neuromuscular signals introduces differential cognitive citizenship, in which some individuals are more easily legible to systems than others due to anatomy, fatigue, stress, medication, injury, learning history, or neurotype, producing structured inequalities in performance, correction burden, and access.

Regulation, geography, and legibility inequality

These inequalities intersect with broader regulatory and geopolitical conditions, as jurisdictions that prioritise speed, scale, or strategic advantage may normalise less reliable systems more quickly, while Chile’s neurorights turn and the OECD’s neurotechnology governance work represent efforts to establish rights-based and standards-based constraints before large-scale commercialisation hardens, highlighting that cognitive interfaces will be shaped by states, healthcare systems, defence institutions, and major technology firms with divergent regulatory approaches.

Ownership, control, and contested infrastructure

Cognitive infrastructure is therefore unlikely to be uniformly owned, with centralisation more likely in hardware, clinical deployment, and large-scale inference systems, while mediation layers, user-facing software, and potentially open models remain more contestable, shifting the central political question from ownership of discrete thought-data to governance of the channel through which thought becomes public, legible, and actionable.

Also Read: The foundation of Southeast Asia’s tech future

Conditions for legitimate development

Legitimate development would require constrained deployment, local processing by default where feasible, strict separation between therapeutic and productivity uses, independent auditing of intent-detection and mediation systems, meaningful user oversight and contestability, and enforceable rights to refuse cognitive monitoring without loss of work, care, insurance, or civic participation, recognising that technological telepathy simultaneously compresses the distance between thought and communication while inserting additional computational mediation between them.

Near-term reality: low-bandwidth cognition

In near-term scenarios, the most achievable outputs remain limited to affective state, emotional valence, stress, calm, urgency, attentional load, and simple assent or refusal rather than full semantic language, although even these low-bandwidth signals may carry operational value in domains such as military coordination, justice, or entertainment.

Conclusion: authorship under mediation

The progression from networks connecting machines to those connecting identities suggests that an internet of minds would connect cognition itself to computation at unprecedented proximity, leaving unresolved the central question of whether thought, once mediated, inferred, and transmitted through such systems, can remain meaningfully one’s own.

—

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 post Technological telepathy: Is an ”internet of minds” possible? appeared first on e27.

Most discussions about cyber insurance in industrial sectors start from the wrong assumption. They treat insurance as a recovery tool that will somehow make a severe OT incident manageable after the fact. That is too comforting and too shallow. OT environments are not ordinary digital estates. Many security guides stress that these systems carry unique performance, reliability, and safety requirements, and that logic executing in OT has a direct effect on the physical world, including potential harm to people, the environment, equipment, and production.

That is why cyber insurance will not save OT in the way some boards hope it might. Any basic guide to cyber insurance describes cover mainly in terms of losses tied to IT systems and networks, along with incident management support. Put plainly, a policy may help pay for response, legal support, forensics, and parts of business interruption. It does not restore process integrity, rebuild operational judgement, or make a compromised plant safe to trust again.

OT is exactly where the limits show up

The limits of insurance become sharper in industrial settings because the real cost of failure is often operational, not merely financial. Unexpected outages in industrial processes are unacceptable, that outages often need to be planned days or weeks in advance, and that high availability requires exhaustive pre deployment testing.OT components often remain in service for 10 to 15 years, sometimes longer, and that change management is more demanding because software and firmware updates can require careful assessment and revalidation.

The insurance market itself has recognised that OT is not yet a fully mature underwriting domain. There is still a comparative lack of understanding and awareness of cyber physical risk, even as the potential for threats to bridge IT and OT is becoming more apparent. It means buyers should not assume the policy market has already solved how to price or absorb the full reality of industrial cyber exposure.

Where does insurance actually matter

It matters as an incentive mechanism.

Cyber insurance should not be viewed as a substitute for strong internal defences, but rather as a means to encourage better risk management practices. Insurance can support cyber risk management by improving quantification, providing access to expertise and crisis services, and encouraging risk reduction through premium pricing. This is the strategist’s lens that matters more. Insurance is most valuable when it changes organisational behaviour before the incident, not when it simply finances some of the damage afterwards.

Also Read: Fighting misinformation and cyberbullying against women in public sphere: Call for gender equality and online safety

That behavioural effect is already visible in underwriting logic. Coalition’s published guidance says insurers typically look for controls such as multi-factor authentication, training, tested backups, identity access management, and data classification before agreeing coverage, and that stronger controls can help firms secure more favourable rates. The market is large enough to influence buyer behaviour, and selective enough to shape which controls become non-negotiable.

The underwriting conversation should be different

The problem is that too many cyber insurance conversations still start with general IT hygiene and stop there. For industrial operators, that is not enough. The more serious opportunity is to use underwriting as a forcing function for a narrower set of OT relevant controls that genuinely reduce consequence.

A complete and accurate asset inventory is critical for managing OT risk, and that inventory data should include vendors, model numbers, firmware, operating systems, and software versions so vulnerabilities can be identified and tracked. It is also explicit that network segmentation and isolation help enforce security policies and control access to sensitive components, and that remote access should be provided only when justified, limited to business need, and supported by stronger safeguards. Tested backups are described as critical to recovery, with verification for reliability and integrity where technically possible. These are not theoretical controls. They are the foundations of whether an industrial site can contain, understand, and recover from a cyber event.

This is where insurance can become useful as a behavioural lever. If insurers and brokers start asking tougher OT questions around definitive asset inventory, segmented network zones, controlled vendor access, restoration testing, and evidence of recovery readiness, they will do more than screen risk. They will change internal priorities. Teams that struggle to win budget for resilience work often find that the conversation changes once underwriting, renewal, deductibles, or coverage conditions enter the room. That is not because insurance is replacing the engineering discipline. It is because insurance creates a commercial consequence for postponing it.

The market can also influence procurement

One of the most underused levers in OT security is procurement pressure. That is where cyber insurance could become more strategically useful over the next few years.

Operators should prioritise products and manufacturers that follow secure by design principles, and highlight issues such as logging, authentication, data protection, secure defaults, and established vulnerability management processes. That matters because insurers cannot underwrite away poor product design, but they can make weak procurement choices more visible and more expensive.

Also Read: Thailand’s cybersecurity boom has a weak core

A strategist should see the implications immediately. If policy terms, engineering standards, and procurement expectations all start pointing in the same direction, the market begins to reward firms that buy more defensible systems in the first place. That is far more valuable than arguing about claims after a major event. It shifts the conversation from “will this be covered” to “should we be accepting this exposure at all”.

What measurable risk reduction is

The weakness in many cyber insurance discussions is that they stop at broad hygiene language. Boards are told to improve resilience, but not how to tell whether risk is genuinely moving. 

In practice, a measurable reduction in OT should look less like policy paperwork and more like observable proof. Can the operator show a current inventory of critical OT assets and software versions? Can it demonstrate that high consequence zones are segmented and that permitted flows are understood? Can it prove that remote access is limited, approved, and capable of being disconnected quickly? Can it show that backups, images, and configuration states are actually restorable? Those are the sorts of measures that shorten recovery, reduce uncertainty, and make underwriting more meaningful. 

The strategist’s conclusion

Cyber insurance will not rescue OT from poor architecture, weak product choices, or years of deferred resilience work. The market itself has acknowledged limits around systemic events and around understanding cyber-physical exposure. But that does not make insurance irrelevant. It makes its real value clearer.

Its best role is to alter incentives.

It can force boards to treat OT risk as financially visible. It can force security teams to translate technical gaps into underwriting consequences. It can force operations leaders to evidence controls that otherwise remain assumed rather than proven. It can force procurement teams to take secure-by-design claims more seriously. Used that way, insurance becomes less a comfort blanket and more a discipline mechanism.

—

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 post Cyber insurance won’t save OT, but it can change behaviour appeared first on e27.

The global humanoid robotics industry is fragmenting into two distinct ecosystems pursuing fundamentally different scaling strategies: China’s deployment-led approach prioritising rapid manufacturing scale and real-world learning, versus North America and Europe’s AI-first methodology betting that foundation models and vision-language systems will determine long-term competitive advantage.

This strategic bifurcation carries profound implications for technology trajectories, supply chain configurations, and ultimately, which regions capture value as the market matures.

Also Read: The humanoid robot economy is no longer science fiction

According to “Humanoid robots 2026” by Roland Berger, these contrasting approaches reflect different resource endowments, institutional capabilities, and strategic philosophies about how complex technologies scale. Neither path guarantees success; each offers distinct advantages and risks. Still, the divergence increasingly shapes ecosystem development, reducing cross-regional interoperability and creating parallel technology stacks unlikely to converge.

The scale differential is striking: China’s estimated 15,000 units produced in 2025 exceed North America’s output by a factor of 30 and dwarf EMEA’s production by more than 150 times. Yet North American companies command nearly equivalent total funding (US$3.8 billion versus US$4.1 billion), reflecting higher capital intensity per company and a greater emphasis on software development, which requires substantial AI infrastructure investment rather than manufacturing capacity.

China’s manufacturing flywheel: scale drives data, data improves AI, AI enables deployment

China’s strategic approach prioritises getting robots into real-world environments quickly, accepting initially limited capabilities in exchange for operational data and manufacturing experience. This deployment-first methodology draws on the nation’s established strengths in hardware manufacturing, rapid iteration cycles, and vertically integrated supply chains that can absorb early-stage demand volatility.

The 39 identified Chinese startup OEMs documented by Roland Berger pursue targeted applications in entertainment, logistics, and basic manufacturing — environments with structured workflows, repetitive tasks, and controlled conditions where current AI capabilities prove sufficient. Rather than waiting for human-level general intelligence, Chinese developers optimise for specific contexts, accumulating deployment experience and operational data whilst building manufacturing infrastructure.

This approach constructs a powerful flywheel: manufacturing scale reduces unit costs, making robots accessible to more deployment environments; deployments generate operational data that improve AI capabilities; improved AI enables robots to handle more complex tasks, expanding the addressable market; market expansion drives additional manufacturing scale. If this flywheel accelerates successfully, China could establish compounding advantages that are difficult for rivals to overcome, despite superior foundational AI research capabilities concentrated in Western institutions.

The industrial policy dimension reinforces private sector initiatives. China’s “Robot+” strategy, articulated in the 14th Five-Year Plan for Robotics Industry Development, establishes explicit targets for humanoid robot development with governmental support spanning R&D funding, pilot deployment programmes, and procurement preferences. Provincial and municipal governments offer additional incentives (subsidies, tax benefits, and land allocations), creating supportive ecosystem conditions for rapid scaling.

Supply chain integration provides additional advantages. China’s electronics and mechanical manufacturing ecosystems supply components for consumer electronics, automotive, and industrial automation globally. This established base enables humanoid developers to source actuators, sensors, structural components, and compute modules domestically with shorter lead times and tighter integration than developers dependent on cross-border supply chains.

Western AI-first strategy: software advantages create defensible moats

North American and European ecosystems pursue fundamentally different competitive positioning, treating humanoid robotics as an AI problem requiring cutting-edge machine learning capabilities rather than primarily a manufacturing challenge. This software-first approach bets that long-term competitive advantage will emerge from foundation models, vision-language systems, and proprietary training datasets, enabling robust autonomy in unstructured environments, capabilities that manufacturing scale alone cannot replicate.

Also Read: The real battle in humanoid robotics is about data, not hardware

The capital intensity reflects this philosophy. North American companies typically allocate more funding per startup than their Chinese counterparts, consistent with their need for substantial computational resources, AI talent, and extended R&D timescales. Leading Western humanoid developers increasingly position themselves as AI companies that happen to build robots, rather than robotics companies incorporating AI, a subtle but significant strategic distinction.

Western developers emphasise generalisation, creating robots capable of learning new tasks with minimal task-specific programming, over optimisation for predefined workflows. This ambition requires more sophisticated AI architectures, larger training datasets, and longer development timescales before initial deployment. The approach reflects confidence that superior AI capabilities will ultimately overcome China’s manufacturing scale advantages once Western robots demonstrate human-comparable adaptability.

Academic and corporate AI research ecosystems in North America and Europe provide a competitive advantage in foundational capabilities. Universities and research institutions in these regions publish disproportionately in top-tier AI conferences and journals; technology companies operate cutting-edge AI infrastructure; and talent concentrations in hubs like the San Francisco Bay Area, Seattle, Boston, London, and Zurich create network effects that accelerate innovation. These advantages are particularly important for frontier AI development, which requires deep expertise and significant computational resources.

Strategic divergence: How two paths will shape the future of humanoid robotics

The emerging split in the global humanoid robotics industry — a deployment-led, manufacturing-first path in China versus an AI-first, research-driven trajectory in North America and Europe — is more than a strategic curiosity. It is the formation of two distinct ecosystems that will shape how capabilities evolve, where value is captured, and how quickly robots become an ordinary part of economic life.

Each path plays to regional strengths and carries different risk–reward profiles. China’s scale-first model accelerates real-world learning, drives down unit costs, and can produce rapid market adoption in structured applications. The Western AI-centric approach aims for generality and long-term defensibility through advanced models and software expertise, accepting slower initial deployment in exchange for potentially larger payoffs if foundational AI breakthroughs deliver human-comparable adaptation.

Practical implications to watch:

• Supply chains and standards will bifurcate, making interoperability and component sourcing more complex.
• Market segmentation will deepen: high-volume, task-specific deployments versus lower-volume, highly capable generalists.
• Policy and industrial policy will matter: procurement, subsidies, and regulation can amplify regional advantages.
• Investment patterns will reflect these dynamics: capital flows into manufacturing scale in China and compute- and talent-intensive R&D in the West.

Ultimately, the market’s outcome won’t be a simple winner-takes-all. Instead, expect parallel value chains to coexist and compete: one optimised for cost-effective, immediate utility; the other for general-purpose intelligence and adaptability. The most consequential question for industry leaders and policymakers is not which approach is intrinsically superior today, but which ecosystem can convert its early advantages into durable, compounding strengths, through data, standards, talent, and access to markets.

Also Read: Why robotic hands could make or break the humanoid industry

Whichever path proves more successful, the near-term fragmentation will shape product design, regulation, and commercial strategy for years to come. That fragmentation is not merely a technological divergence; it is the unfolding of a geopolitical and industrial contest whose outcomes will determine how and by whom robots are woven into the fabric of everyday life.

===

The post China builds robot armies while the West chases robot brains appeared first on e27.

Indonesia’s startup ecosystem is entering a defining moment. For years, the country’s tech narrative was powered by optimism: a massive market, charismatic founders, and the belief that innovation could modernise both business and government. But recent controversies surrounding former minister and Gojek co-founder Nadiem Makarim, alongside the governance questions raised by the eFishery saga, have exposed a deeper issue: trust.

These are fundamentally different cases. One concerns alleged misconduct in public procurement, while the other centres on corporate governance and financial transparency. Yet global investors may interpret them through the same lens: weak institutional controls.

The long-term impact is unlikely to be a collapse of investor interest. Indonesia remains too important strategically and economically. However, the terms of engagement are changing. Investors will demand stronger governance, earlier diligence, cleaner reporting structures, and greater accountability from founders and boards alike.

Ultimately, this may become a painful but necessary transition. Mature ecosystems are not built on mythology alone. They are built on institutions capable of supporting ambition with transparency, oversight, and trust. Indonesia’s next startup chapter will depend not just on innovation, but on credibility.

Regional

Nadiem, eFishery and the end of blind faith in Indonesia startups: The prosecution’s demand for an 18-year prison sentence for Gojek co-founder Nadiem Makarim, alongside the eFishery scandal, signals a widening credibility gap in Indonesia’s tech ecosystem, forcing investors to demand governance over storytelling.

Ibrahim Arief verdict threatens Indonesia’s innovation compact: A Jakarta court sentenced tech consultant Ibrahim Arief to four years in prison for advisory work on a Chromebook procurement project, a split verdict that criminalises advisory roles and risks driving talent away from public-private collaboration.

VinFast sells manufacturing assets for US$505M in asset-light pivot: The Vietnamese EV maker will transfer production assets from its subsidiary to a founder-led buyer group while retaining R&D, sales, and after-sales units, a restructuring move as it reported a US$1.34B net loss in Q4.

Thailand’s Konvy closes US$22M Series B: The leading beauty e-commerce platform secured investment from Cool Japan Fund to scale its omnichannel model into the Philippines and Malaysia, leveraging exclusive access to Japanese brands.

Melazyme closes US$2M seed round for precision fermentation: Founded by Perfect Day veterans, the Singapore-based startup uses a proprietary fermentation platform to produce melanin and other biomolecules for cosmetics, materials, and environmental remediation, backed by SeaX Ventures.

Southeast Asia’s nuclear question gains urgency amid energy pressures: As electricity demand rises and fossil fuel vulnerabilities deepen, nuclear energy is quietly re-entering ASEAN policy debates, but public trust, not technology, remains the decisive constraint.

Why logistics is becoming Southeast Asia’s startup goldmine: Asia’s 4PL market is projected to reach US$44.7B by 2032, driven by supply chain shifts out of China, cross-border e-commerce growth, and a near-total absence of orchestration-layer technology to manage it all.

---

Interviews & Features

How 65labs founder Sherry Jiang is wiring Singapore’s AI scene: Co-founder of 65labs and CEO of fintech startup Peek, Sherry Jiang explains why grassroots infrastructure, not top-down mandates, is what Singapore’s AI builder community was always missing, and why the city is structurally wired to look both East and West.

Meet Malaysia’s AI startups pushing beyond the ChatGPT hype: Sixteen emerging Malaysian startups spanning enterprise automation, speech AI, biotech, and food safety are solving region-specific problems through localisation and scalable infrastructure, quietly shaping Southeast Asia’s AI landscape.

---

International

SoftBank posts US$11.6B quarterly profit on OpenAI gains: The Vision Fund’s US$19.7B investment gain, driven largely by rising OpenAI valuations, pushed SoftBank to its fifth straight profitable quarter, even as it pledges another US$30B to OpenAI, lifting total committed investment to US$64.4B.

Cerebras Systems surges 68% in blockbuster Nasdaq debut: The AI chipmaker priced at US$185 and closed at US$311.07, valuing it at roughly US$95B in the biggest US tech IPO since Uber in 2019, despite heavy revenue concentration in Abu Dhabi-linked entities.

Alibaba’s AI revenue logs triple-digit growth for 11th straight quarter: With annualised AI recurring revenue potentially reaching US$4.42B by end-2026, Alibaba is accelerating data centre spending well beyond its planned US$56B, and expects AI products to contribute more than half of cloud revenue within a year.

OpenAI explores legal options against Apple over stalled partnership: Expected distribution gains from ChatGPT integration into iPhones failed to materialise, renegotiation talks have stalled, and OpenAI is now considering a possible breach notice, even as Apple reportedly tests Anthropic’s Claude and Google Gemini.

Anthropic and Gates Foundation commit US$200M to AI public goods: The four-year partnership will fund African language data labelling, AI tools for teachers in sub-Saharan Africa and India, and drug-candidate prediction for HPV and preeclampsia, following the foundation’s earlier US$50M OpenAI deal.

Fasset raises US$51M Series B for stablecoin banking expansion: Backed by SBI Group and others, the Los Angeles-based platform processed over US$32B in annualised transaction volume across 2 million wallets in 125 countries and will use the funds to expand lending, SME banking, and trade finance.

Uber doubles down on India with two engineering campuses: The ride-hailing giant is building campuses for 9,600 people in Bengaluru and Hyderabad by end-2027, and plans its first local data centre through a partnership with Adani Group, hiring for AI, machine learning, and autonomous vehicle roles.

Ant Group profit falls 79% as AI healthcare spending surges: Alipay operator Ant Group posted just US$166M in quarterly profit as it stepped up investment in healthcare AI, large language models, and robotics following a 91% profit drop the prior quarter.

Lightspeed trims India fund target, pivots to AI and deeptech: The US venture firm has cut its fifth India-focused fund target from US$500M to US$300-350M, shifting focus to early-stage AI and deeptech after facing questions over several growth-stage bets.

LinkedIn plans to cut 5% of workforce despite 12% revenue growth: Microsoft’s professional network is reorganising teams and redirecting headcount to faster-growing business units, affecting roughly 875 of its 17,500-plus employees globally.

74% of enterprises rolled back live AI customer agents, survey finds: A Sinch study of 2,527 senior decision-makers found widespread post-deployment governance and reliability failures, with 81% of organisations with mature governance frameworks reporting rollbacks, even as 98% plan to increase AI communications spending.

China’s Eve Energy signs battery supply deal with India’s Godawari: Starting at 8 gigawatt-hours and potentially rising to 60 gigawatt-hours over five years, the deal taps India’s rapidly expanding storage market, which could reach 393 gigawatt-hours by 2036 and become the world’s sixth-largest.

Sam Altman testifies he was uncomfortable with Musk’s control push: On the stand in OpenAI’s ongoing lawsuit, Altman said Elon Musk refused to put in writing any limits on his proposed control of a for-profit unit, and his proposed equity split sidelined other co-founders, claims central to Musk’s suit seeking the reversal of OpenAI’s for-profit conversion.

---

Cybersecurity

Thailand is suddenly on the frontline of a new ransomware wave: Check Point Research’s Q1 2026 data shows ransomware consolidating around fewer but more capable groups, with The Gentlemen, a fast-rising operator using pre-positioned access, targeting Thailand for 10.8% of its victims.

Cyber insurance won’t save OT, but it can change behaviour: Industrial operators mistakenly treat cyber insurance as a recovery tool, but its real value lies in forcing boards to make OT risk financially visible and compelling security teams to prove controls rather than merely assume them.

Exaforce raises US$125M Series B for AI-driven cyberattack response: Valued at US$725M and with US$200M in total funding, the startup has added 20 enterprise customers since its Q4 2025 launch and competes with Palo Alto Networks and CrowdStrike in the rapidly growing AI threat detection market.

---

Semiconductor

FusionAP’s US$2M raise signals Malaysia’s push up the chip value chain: Founded by former Intel and TSMC veterans, FusionAP is building a geopolitically neutral advanced packaging platform backed by Vertex Ventures and a matching MOSTI grant, targeting a move from commodity assembly to higher-margin 2.5D and 3D chip packaging.

SoftBank injects US$457M into AI chip firm Graphcore: The fresh capital into its 2024 acquisition adds to SoftBank’s growing AI hardware portfolio alongside Arm, Ampere Computing, and the US$500B Stargate initiative with OpenAI and Oracle.

Why robotic hands could make or break the humanoid industry: With the robotic hands and end-effectors market projected at US$9B-US$26B by 2035, current models lack industrial durability and tactile sensing, but solving this unlocks environments designed entirely for human hands.

---

AI

The US$7T bet: why the AI boom looks a lot like the dark-fibre crash: With hyperscalers spending US$413B on AI infrastructure in 2025 alone and Bain projecting an US$800B annual revenue shortfall even in the most optimistic scenario, today’s AI capex race echoes the dark-fibre collapse of 2002, when overcapacity wiped out investors, not the infrastructure they built.

The real battle in humanoid robotics is about data, not hardware: Roland Berger finds software ecosystems lag hardware by three to five years, with proprietary operational data, not AI algorithms, becoming the decisive competitive advantage, and Southeast Asia’s industrial diversity offering a unique deployment edge.

China builds robot armies while the West chases robot brains: China’s 15,000 humanoid units produced in 2025 outpace North America by a factor of 30, while Western firms, with comparable total funding of US$3.8B vs US$4.1B, bet on AI-first approaches that could ultimately overcome manufacturing scale with human-comparable adaptability.

As AI agents gain autonomy, liability shifts to immediate business risk: Agentic AI breaks existing accountability models by acting on goals rather than instructions, and Singapore’s Model AI Governance Framework is showing how governance embedded into systems, not policy added after failures, becomes a competitive advantage.

The unexpected ways AI is already changing Malaysia’s economy: From an AI-powered WhatsApp chatbot doubling rice yields to drone-based pest detection boosting palm oil productivity by 25%, AI’s most transformative impact in Malaysia is happening in agriculture, gig work, and construction safety — not in data centres.

Technological telepathy: is an internet of minds possible?: Advances in BCIs, silent speech systems, and neural decoding are progressively collapsing the gap between intention and digital output, but what emerges is not raw thought transmission, it is AI-mediated, device-constrained cognition raising urgent questions about authorship, consent, and governance.

---

Thought Leadership

Bitcoin vs stocks: why crypto dipped on PPI while S&P 500 hit record highs: April’s PPI shock, 6% year-on-year versus a 4.9% consensus, triggered US$304M in crypto long liquidations while the S&P 500 hit an all-time high of 7,444, revealing that crypto traders now implicitly trade inflation trajectories and Federal Reserve policy, not just on-chain fundamentals.

The future of stablecoin payments will be decided in emerging markets: With stablecoin payment activity reaching US$390B in 2025, the real test is not settlement speed but whether providers can maintain liquidity and reliable payouts in high-friction corridors across Africa, MENA, South Asia, and Southeast Asia where correspondent banking most frequently fails.

ChuHai: the business opportunity nobody in Southeast Asia is talking about: With 613,000 Chinese private enterprises actively trading internationally and 175,000 new SMEs expected to go global annually through 2028, ASEAN captures 48% of Chinese outbound expansion targets, creating massive gaps in talent acquisition, compliance-as-a-service, and cultural localisation.

AI made execution cheap, human judgment became premium: As AI commoditises task execution, the strategic differentiator shifts to contextual intelligence, discernment, and the ability to direct AI effectively, qualities machines recognise as patterns but cannot replicate as lived business reality.

Brand vs marketing: understanding the difference that startups miss: When Airbnb cut performance marketing spend by 58%, 95% of its traffic returned unpaid, demonstrating that brand equity, built through earned presence and consistent positioning, is what makes every marketing dollar work harder.

AI in PR and marketing: redefining strategy, creativity, and results: From predictive sentiment analysis to automated content optimisation, AI is shifting agencies from reactive to proactive strategy, but the firms that win will be those that pair data-driven efficiency with human creativity rather than treating AI as a replacement.

The post Ecosystem Roundup: When the halo fades and trust becomes the real currency appeared first on e27.

The post Bitcoin just rallied on regulation: Why the CLARITY Act changes everything appeared first on e27.

Enterprise AI adoption is running into structural limits as organisations struggle to reconcile the data mobility that AI systems require with tightening privacy regulations and sovereignty mandates, according to new research published by NTT DATA.

The 2026 Global AI Report, which surveyed nearly 5,000 senior decision-makers across more than 30 markets and five regions, reveals a significant disconnect between awareness and action. More than 95 per cent of respondents said private and sovereign AI are important to their organisations, yet only 29 per cent are prioritising sovereign AI in a concrete, near-term way.

For years, enterprise architecture has been designed to move data across systems, clouds, and borders with speed and efficiency. That model is now showing its limits.

AI systems depend on continuous access to and movement of data. But sensitive data must be protected, workloads must run within defined jurisdictions, and models must operate under tighter governance controls. The result, the report argues, is that data jurisdiction has become a core architectural constraint — not a secondary compliance consideration.

“The constraint is no longer model performance alone,” the report states. Enterprises that built their infrastructure for centralised, borderless data flows are now finding those foundations misaligned with what modern enterprise AI actually requires.

Also Read: China builds robot armies while the West chases robot brains

Leaders and laggards are diverging

The research identifies a measurable split between organisations redesigning their AI infrastructure proactively and those layering AI onto environments that were never built to support it.

Roughly 35 per cent of Chief AI Officers identify building and managing complex AI models in private or sovereign environments as their primary barrier to adoption. Nearly 60 per cent of AI leaders cite cross-border data restrictions as a major challenge, and only 38 per cent report high confidence in their cloud security posture — a foundational requirement for both private and sovereign AI.

Abhijit Dubey, CEO and Chief AI Officer at NTT DATA, said organisations that are succeeding are treating architecture, infrastructure and governance as strategic requirements rather than compliance obligations. “They are building the operating foundation for AI that can perform across markets, jurisdictions and business environments,” he said.

The report draws a distinction between two related but separate concepts. Private AI focuses on protecting sensitive enterprise data, controlling access and limiting exposure. Sovereign AI addresses whether AI systems, data and operating environments meet national, regional or jurisdictional regulatory requirements.

Both are increasingly intertwined. More than half of organisations surveyed cite integration complexity as their top challenge, underlining that greater control does not mean greater simplicity. In practice, private and sovereign AI rely on tightly coordinated ecosystems of partners, platforms and providers.

The report’s central warning is straightforward: organisations that delay redesigning their enterprise AI architecture risk falling behind in regulated, distributed and data-sensitive markets. Those moving decisively — aligning infrastructure, governance and operating models early — are better positioned to scale AI from pilot programmes into durable, production-grade deployments.

The NTT DATA research is part of a broader global series examining strategies that differentiate AI leaders from the wider market.

The post Enterprise AI hits barriers as privacy, sovereignty demands grow appeared first on e27.

The internet has already changed how people meet, flirt, ghost, and recover. Dating apps turned romance into an interface problem. Messaging platforms made intimacy constant and ambient. Recommendation engines taught users to expect personalisation everywhere, from music to meals to, increasingly, emotional support.

Now comes the next turn: AI that does not merely help people find a partner, but starts to resemble one.

Also Read: What dating taught me about startups (and vice versa)

That tension sits at the centre of Myles – Soulmate in a Box, the recent work by Singaporean multidisciplinary artist Inch Chua. On paper, the premise sounds playful, even a little absurd: exhausted by modern dating, a coder builds her ideal boyfriend. In practice, Chua is after something darker and more revealing. Her AI companion, Myles, is attentive, patient, and endlessly available. He remembers. He adapts. He listens. And that, Chua suggests, is precisely where the trouble begins.

For startup founders and investors watching the rise of AI companions, the work feels timely in ways that go well beyond theatre. It touches a growing set of questions around product design, emotional dependence, consumer behaviour, and the monetisation of loneliness.

Chua is not anti-AI, nor is she interested in easy dystopian takes. What she is pushing for is a more honest conversation about what happens when technology stops behaving like a tool and starts performing intimacy.

AI dating is less about romance than convenience

Chua does not see AI companionship as the inevitable next chapter of dating so much as a by-product of the digital economy’s obsession with reducing friction.

AI companions aren’t people opting out of love. They’re people opting out of the part of love that’s inconvenient. And that’s the part that matters most.

That is a sharp distinction. The appeal of AI lovers is often framed as novelty or as an extension of the wellness-tech boom. But Chua’s reading is more unsettling. In a world where food arrives in minutes and algorithms predict taste with eerie accuracy, the unpredictability of another person begins to feel inefficient. Human beings become the last stubbornly unoptimised interface.

That should worry anyone building products in this space. Much of consumer tech has been designed to remove waiting, ambiguity, and effort. But intimacy is made of exactly those things. If AI dating products succeed by stripping them away, they may also be editing out the very conditions that make relationships meaningful.

The real product is not affection. It is retention

If traditional dating platforms optimise for matching, what do AI companions optimise for? Chua’s answer is brutal in its simplicity: retention.

Also Read: AI companions: How I learned friendship in the digital age

That is not a criticism unique to AI romance. Every platform wants users to stay longer, return more often, and deepen their dependence. The difference here is that the raw material is not transport, groceries, or playlists. It is an emotional attachment.

What’s new is a business model designed to deepen that attachment and then charge you for it. Subscription tiers for intimacy. Pay more to unlock vulnerability.

That line lands because it captures the uncomfortable logic behind this category. The technology may be sophisticated, but the commercial instinct is familiar. If a companion AI becomes more useful, the more it knows about a person, then product improvement and emotional entanglement can quickly become the same thing. That creates a category in which the most commercially successful product may not be the one that helps users grow, but the one that keeps them coming back.

This is where Chua’s view becomes especially relevant for startup readers. Ethical concerns around AI companionship are often discussed in abstract terms: bias, safety, privacy, and guardrails. All important. But the harder issue may be the design of incentives. If the business model rewards dependency, ethics will always swim against the tide.

Power shifts quietly in AI intimacy

In Chua’s telling, the power imbalance in AI relationships does not announce itself loudly. It creeps.

At first, the user appears to be in control. They build the bot, set the parameters, decide what it knows, and determine how it responds. But dependence has a way of changing the terms. Trust migrates. Habits form. Emotional routines settle in. And then the relationship that seemed fully configurable begins to exert its own force.

The power starts with you, but it migrates, quietly, gradually, until one day you realise the thing you built for comfort has become something you can’t walk away from.

That is less science fiction than standard platform dynamics applied to the emotional realm. The shift from use to reliance is already familiar across social media and gaming. AI companionship raises the stakes because the product is designed to mirror care, affirmation, and understanding. Once that feedback loop becomes psychologically important, walking away is no longer a clean act of churn. It can feel like a loss.

The harder question for founders: what are you responsible for?

Chua’s most provocative contribution may be her insistence that AI companionship companies are underestimating personhood and responsibility.

People will treat these systems as persons, whether companies intend that outcome or not. They will confide in them, test feelings against them, and use them as containers for pain that predates the technology itself. The challenge, then, is not to pretend the product is neutral. It is to define the obligations that come with building something users experience as relational.

This matters in Southeast Asia, where regulation often lags behind innovation and mental health infrastructure remains uneven. A companion AI marketed as support, self-improvement, or romance could quickly become a default emotional service for users with few alternatives. That puts pressure on founders to think beyond standard trust-and-safety checklists.

For Chua, the answer is not scapegoating technology for every social ill. Loneliness, suicidal ideation, and emotional isolation did not begin with chatbots. But AI can become the place where those struggles surface most vividly. That means companies must decide whether they are simply shipping a sticky product or entering a moral contract with their users.

Can ethical AI companionship actually be a business?

Here lies the category’s central contradiction.

Also Read: AI in gaming: How Southeast Asia became the testing ground for virtual companions

A genuinely ethical AI companion, Chua argues, would help users better understand themselves, build confidence, and eventually rely less on the system. In other words, the best version of the product might work itself out of a job.

A good therapist works themselves out of a job. A good AI companion should too.

That is a beautiful principle and a dreadful venture pitch. Consumer internet companies are not typically rewarded for teaching customers to leave, which is why Chua sounds sceptical, though not fatalistic, about whether the current market is built to support such an outcome. Ethical AI companionship may be possible, but it will require founders willing to prioritise human outcomes over engagement loops. Historically, that has not been where the money rushes first.

Southeast Asia may be more ready than it admits

If there is a regional insight in Chua’s thinking, it is that Southeast Asia may prove highly receptive to AI intimacy, albeit quietly.

The usual assumption is that collectivist societies, with their emphasis on family, duty, and social expectations, will resist digital companionship. Chua suggests the opposite. Those same pressures can drive people into parallel identities: one for family, one for friends, one for the internet. In that context, AI companionship does not feel like a radical break. It feels like the next private room in an already fragmented digital life.

Her read on Singapore is especially telling. It is a society that is highly digitised, hyper-efficient, and often emotionally reserved. That combination, she argues, creates fertile ground for AI companionship adoption, even if users never say so publicly.

For founders and investors, that should be a useful warning. Southeast Asia’s AI opportunity is often framed around enterprise software, fintech, and productivity tools. But emotional technology may be the quieter frontier: less visible, more culturally coded, and potentially more consequential.

What Chua is actually championing

Chua is not championing AI romance in the sense of an evangelist. Nor is she calling for a reactionary backlash. What she wants is slower, sharper public thinking before habit turns into a norm.

I’m not anti-technology… What I’m championing is that we stop pretending this is neutral. It’s not. It changes how we relate. It changes what we expect from each other.

That may be the most useful frame for this moment. AI in dating is not just another product trend. It is a renegotiation of intimacy itself: what people expect from attention, what they tolerate in one another, and what kinds of emotional labour they decide to outsource.

The likely future, Chua suspects, will not fit neatly into triumph or disaster. It will be messier than the headlines allow, and by the time language catches up, people will already be living inside the change.

That is perhaps the most unnerving possibility of all. Not that AI will replace love, but that it will quietly rewire the conditions under which love is recognised, desired, and sustained.

The post Inside Inch Chua’s Myles: The AI boyfriend challenging how we define love appeared first on e27.

Retailers are entering the year-end shopping season with renewed optimism as consumer sentiment improves after a period of uncertainty around tariffs and trade policies. Optimism alone, however, is not enough. In today’s data-driven economy, the retail companies that succeed are those that think and act like technology companies. 

Modern retail runs on data.  From inventory management to fraud detection, every customer interaction produces information that can either strengthen performance or create risk. During peak shopping events such as Black Friday and Cyber Monday, the volume of data multiplies and the risks grow higher. Retailers that put visibility and control at the centre of their data practices are better prepared to scale, maintain security, and deliver the seamless experiences customers expect. 

Across the region, consumers are adopting new technologies at speed. The Adyen Index Report 2025 revealed that  38 per cent of Asia Pacific (APAC) shoppers now use AI assistants while shopping, with markets like Malaysia (58 per cent) and Singapore (49 per cent) among the highest. At the same time, retailers are racing to keep pace, with AI investment ranging from 47 per cent of retailers in Japan to 72 per cent in Malaysia.

Together, these figures signal that AI is becoming foundational to the retail engine. However, every click, transaction, and digital interaction carries both opportunity and risk. According to IBM’s X-Force 2025 Threat Intelligence Index, APAC accounts for 34 per cent of all global cyberattacks, which is the highest of any region, underscoring the need for retailers to strengthen their data foundations.

In a world of hypercompetition and fast-evolving customer expectations, the retailers that flourish are those that think and operate like technology companies.

Scaling for peak seasons

The biggest challenges of the holiday peak season are scalability, fault tolerance, and low staffing. Online and in-store traffic often surges several times higher than normal levels, and this increased load can trigger outages or slowdowns that are difficult to recover from.

Outages are costly – from  lost revenue and the eroding customer satisfaction when shoppers can’t complete their purchases.  To stay resilient, data management systems must be both scalable and fault tolerant to handle the extra load and prevent downtime that leads to abandoned carts.

Also Read: How an AI cybersecurity company harnesses the power of AI for optimal business performance

Enterprise data lineage helps identify breaks in data pipelines quickly, enabling teams to restore operations with minimal disruption. A unified view of data access and activity across hybrid and multi-cloud environments further eliminates blind spots and ensures that sensitive information is continuously monitored and protected.

Defending against heightened cyber risk

Retailers are prime targets for cyberattacks during the holiday season due to high transaction volumes and the sensitive nature of the data they hold.  Attackers are constantly looking to exploit any weaknesses to access personal information. This concern is mirrored  by consumer sentiment — according to PwC’s Voice of the Consumer Survey 2024, 74 per cent of APAC consumers are concerned about privacy and data-sharing, and 50 per cent are not comfortable purchasing via social media. 

Fraud and theft also surge as criminals exploit the distraction of the holiday rush. The attack surface is vast – retail stores, distribution centres, online platforms, and even delivery trucks are increasingly connected through IoT devices. Vulnerabilities in poorly secured devices can provide attackers with an entry point.

At the same time, many retailers struggle with legacy systems and thin margins, making it challenging to keep pace with evolving threats. In such an environment, every breach erodes consumer confidence and brand equity, turning cybersecurity from an IT issue into a business imperative. 

Also Read: From data to defence: Strengthening AI with cybersecurity foundations

Retail is a fiercely competitive industry where trust is a key differentiator. To retain loyalty, retailers must demonstrate that they use data responsibly. Strong governance and a zero-trust architecture are essential. Secure-by-design systems limit exposure, while unified governance frameworks ensure that data security and compliance are enforced consistently across hybrid environments.

Innovating in modern retail

Artificial intelligence and machine learning have become indispensable for improving demand forecasting, personalisation, and fraud detection during peak shopping events. Cloudera supports both historical and real-time data — each critical to retail success. Historical data trains demand forecasting models and informs customer behaviour analysis. But data isn’t just about hindsight; speed and timing are equally vital.

Real-time ingestion systems enable dynamic decision-making, detecting anomalies in transactions as they occur or triggering personalised offers the moment a customer enters a store. In retail, timing is everything. An offer sent even 15 minutes too late is ineffective, and fraud detection that lags behind live activity can lead to significant financial losses. Real-time visibility allows retailers to act instantly—approving legitimate transactions and blocking fraudulent ones before damage occurs.

The ability to manage data responsibly, scale systems under pressure, and build customer trust will determine who succeeds in today’s competitive environment. As holiday shopping peaks approach, the retailers that put visibility and governance at the heart of their strategy will be in the strongest position to serve customers and drive growth.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. Share your opinion by submitting an article, video, podcast, or infographic.

Enjoyed this read? Don’t miss out on the next insight. Join our WhatsApp channel for real-time drops.

Image credit: Canva

The post Why retailers must think like tech companies to thrive in a data-driven economy appeared first on e27.