Singapore as AI-Era Globalization Bridgehead: GCS Day 1 Insights on Growth & Applied AI

Global Growth Summit: Singapore as a Globalization Bridgehead

White Whale Outbound founder and CEO Wei Fangdan opened the conference noting that Chinese enterprise globalization has moved from exploration to deep cultivation, becoming a "must-answer question." The focus has shifted from opportunity-driven to capability-driven, from pure efficiency to balancing efficiency, resilience, and security. AI is lowering globalization barriers and accelerating the process.

Singapore is emerging as a key node where Eastern and Western AI forces converge. Western AI giants (OpenAI, NVIDIA, Google DeepMind, Anthropic) and Chinese tech leaders (Alibaba Cloud, Tencent, ByteDance) are both using Singapore as a strategic hub connecting Southeast Asia and global markets. Its geography, multinational corporate presence, and AI ecosystem make it a nexus for technology, talent, capital, and markets.

Three Keywords for Singapore's Value

Connectivity & Resonance: Singapore's core advantage is connecting global resources. Beyond venues and hardware, it is deepening partnerships with Chinese industry associations and alliances, transforming MICE events from one-off gatherings into platforms linking industrial resources and international markets.

Ecosystem Innovation: Moving from product export to model, content, and ecosystem export. Singapore aims to be the launchpad for new models, blending business meetings with premium entertainment (F1, music festivals) under the "Bleisure" trend, enabling B2B and B2C dual-track ecosystems for Chinese IP and brands.

Sustainable Symbiosis: Providing stable, transparent business and legal environments for long-term enterprise development, while advancing sustainable development and AI capabilities so companies can validate products and models before expanding to wider Southeast Asian and global markets.

These three directions form a virtuous cycle: "business attracts meetings, meetings promote business."

Globalization Restructuring in the AI Era

Deng Xin (Cindy) , Associate Dean of Nanyang Business School, observed that technology and infrastructure are erasing geographic distance, while trade and politics are thickening borders. Companies no longer face a binary choice between globalization and deglobalization; globalization itself is being restructured. The core concern has shifted from efficiency to resilience — safety, market distance, regulatory fit, and alternative options now matter as much as cost.

AI is a critical new variable. Unlike past capital-intensive, long-cycle infrastructure revolutions that favored large organizations, AI lowers the barrier for any enterprise to compete globally. Market research, content production, customer service, and localization — once high hurdles — can be dramatically reduced. But as AI becomes ubiquitous, "having AI" will no longer differentiate; the real gap will be how well a company understands markets, cultures, and institutions.

If customers, talent, capital, data, and supply chains are distributed across countries from day one, the enterprise must design its architecture as a global-native company: a multi-layer structure of global core, regional hubs, and local units, with the ability to recombine as regulatory and geopolitical environments change.

Generative AI Content, Monetization & IP

Jaze Phua , founder of GenTube.com, noted that by 2025 roughly 74% of new web content involves AI in creation, leaving only about a quarter purely human-made. This explosion has brought a flood of low-quality "AI slop" that buries quality content and hurts creator revenue.

GenTube splits AI-era creators into GenCreators (original IP creators) and GenTubers (AI-avatar livestreamers), offering monetization via subscriptions, live gifts, and tips. It builds a triangular ecosystem: AiCON for community (creator-audience entry), DEAR.AI for agency and production (serving brand budgets), and GenTuber as the content/traffic core linking creators, audiences, and brands.

Singapore's Startup & Innovation Ecosystem: Opportunities & Pitfalls

James Tan , Managing Partner of Quest Ventures, warned of a dangerous assumption: surviving China's hyper-competitive internet market guarantees easy success in Southeast Asia. Many failures stem from arrogance:

Copy-pasting China standards & speed: Imposing 996 culture causes high turnover and labor-law backlash; applying China's high-growth playbook misreads local markets and users.

Treating Southeast Asia as monolithic: The region is highly fragmented geographically and culturally; teams that refuse to learn local languages/cultures or empower local partners often fail.

Relying on halo effects: Entering with a "bringing advanced productivity to developing regions" attitude ignores world-class local players (Indonesian beauty, Vietnamese game studios, Singaporean fintech) and breeds resentment.

Singapore's value lies in forcing governance upgrades (international-standard equity, IP, compliance frameworks), serving as Asia's only hub integrating top Western engineers, Southeast Asian operators, and Chinese technologists, and providing a transparent regulatory sandbox where business models are validated in a rules-based environment before scaling — earning trust from day one.

Roundtable: Hyper-Localization in a Multipolar World

Helen Wong (ACV Capital): Entering Indonesia requires navigating licenses, certifications, and compliance — adding time and cost. Understanding what products and content truly connect with local consumers is essential.

Elle Liu (Canva Singapore & Malaysia / SEA Strategic Partnerships): Balance standardization and localization like a "cupcake": core product and base capabilities stay unified; registration, payment, pricing, and content channels adapt to local tech ecosystems and user habits. Building relationships with local associations, government bodies, and user communities establishes credibility and leverages advocates for reach.

LAU Wee Kiat (Singapore AI Association CTO): AI is a key cross-market operational tool but cannot replace local relationships and market understanding. The association helps SMEs identify AI application gaps, match solutions/training, and partners with overseas AI associations to facilitate market entry. AI can slash localization costs (e.g., Chinese HQ generates marketing creative, then AI produces Vietnamese/Philippine variants), yet human-to-human relationships remain the foundation of cross-border execution.

Applied AI Summit: Trust, Agents, Sovereign AI, Recruiting, Product Love & Film

Building Trust for the Next Decade

Binghan Yi , Co-founder & CFO of WIDTH, declared: "Last decade compared speed; next decade compares trustworthiness." Two forces drive this: regulation (expanding scope, finer granularity, per-decision accountability, divergent cross-border rules) and generative AI enabling low-cost, scalable fraud (synthetic identities, deepfakes). Compliance workloads are surging.

Regulatory scope now covers cross-border payment/settlement providers, platform fund flows, digital asset trading, even game/social virtual asset top-ups. Review frequency has moved from annual to per-transaction traceability; every decision needs documented rationale. A single weak link can jeopardize licenses, payment channels, app-store eligibility, and personal liability. Divergent jurisdictional rules mean the same business faces different obligation checklists per market — manual handling is impossible; rules must become configurable.

AI is both risk amplifier (fraud at scale) and compliance solution: it can pipeline information extraction, rule matching, reasoning, and disposition recommendations, letting risk teams focus on truly judgment-required cases.

Agent Era Globalization Breakthrough: From Token Economics to High-Quality Delivery

Rock Shi , Google Cloud AI Product Sales Consultant, presented data: average per-request token consumption up 8× YoY (up to 20× in some scenarios). Though token prices fell 98% since 2024, many AI companies see inference costs rise. Goldman Sachs predicts another 24× token usage growth by 2030.

This creates an "AI product paradox" : traditional SaaS enjoys ~80% gross margins because user growth adds near-zero marginal cost; AI products average ~52% gross margin because each inference costs money, and deeper reasoning consumes more tokens. Success can worsen finances.

Scale no longer simply dilutes cost. AI products must control cost at the granular level: abandon "always use the best model" thinking. Different tasks have different requirements; over-using strong reasoning models wastes money and can hurt performance. Companies need scenario-specific evaluation systems, defining the AI's "KPIs," translating them into measurable model metrics, and verifying that the model meets expectations — achieving both task completion and cost control.

Building Enterprise Sovereign AI Strategy

Bruce Yang (Yang Tong) , CEO of Agnes AI, argued that in the Agent era, "renting intelligence" exposes weaknesses: no model ownership, no capability moat, plus cost, API latency, data security, and vertical fit issues. Enterprises are asking how to own their AI capabilities.

Agnes AI helps build a sovereign AI tech stack — not training from scratch, but establishing an autonomous, controllable, continuously evolving stack:

Define model role in business; build evaluation around real scenarios.

Intelligent routing: dynamically allocate tasks across models, vertical tools, and human review based on complexity, risk, latency, cost.

Inference optimization to cut runtime cost; post-training to make models truly understand the business; use Agent execution traces to continuously improve models.

Online learning so models evolve from business feedback.

Agnes provides the full stack from model, eval, routing, inference optimization, post-training, to Agent and online learning — turning AI from "externally rented" to "owned, controlled, iterated" core productivity. Their models serve >1 trillion tokens/day and are opened to enterprise clients.

AI Recruiting's Double Revolution: Blue-Collar to White-Collar, Employer to Job Seeker

Kevin Gao (Gao Yi) , CEO of Arros AI, cited data: LinkedIn receives 11,000 applications/minute in 2026 (+45% YoY), yet 77% of APAC employers struggle to fill roles. AI drives applicant marginal cost to near zero while skyrocketing employer verification/matching costs.

Blue-collar pain point: process length. Frontline workers decide in ~24 hours; traditional web applications are complex, losing ~50% mid-flow, then wait for manual screening/scheduling. Solution: compress apply-screen-schedule into one conversation; let workers with only a phone apply in native language via chat and lock in interview/shift instantly.

White-collar pain point: screening. Candidates use AI for resumes, applications, interviews → volume up, resumes homogenized. Employers deploy AI interviewers/agents → arms race (better detection → better disguise). Solution: redesign hiring signals — verifiable identity/skill layers, upfront AI-use disclosure, retain human review/interview. AI handles initial screening & info synthesis; humans make final judgment.

Beyond Product-Market Fit: Building AI Products Users Truly Love

Felix Wong , Co-founder & CPO of Emoka, noted AI makes prototyping trivial (hours vs weeks). When "build" becomes cheap, "understand" becomes scarce. Emoka breaks user understanding into four layers: Problem, Behavior, Emotion, Commitment .

In early Japan exploration: ~200 interviews, 1,000 surveys, multi-region user tests. They watched what users did and what they'd do for the product , not just what they said. Insights drove concrete decisions: adjust product colors for Japan, offer stronger early-bird perks to long-term followers, users voluntarily joined a Tokyo Shibuya shoot for free.

True insight isn't collecting feedback; it's understanding what users will do for the product. Beyond explicit needs, behavior reveals unexpected opportunities. When Emoka exhibited in Harajuku IP spaces/galleries alongside artists/designers, audiences asked not "What can Emoka do?" but "What world does Emoka belong to?" This reframed Emoka from a robot to a character, art piece, even consumer brand. Users began co-creating that world — one early buyer built a fan page and drew AI-generated comics of daily life with Emoka. Users bring the product into their lives and assign personal meaning.

Conclusion: AI lowers the "Build" barrier but raises the value of "Understand." The next phase isn't about speed of output, but depth of human understanding and building relationships users want to participate in, connect with, and sustain.

From "Generating Video" to "Orchestrating Creativity": AI Film's Next Decade

David Wang , VP Global Distribution at FizzDragon, argued that as visual production democratizes and barriers drop, generation capability is no longer scarce. Scarce resources become: continuous judgment, collaboration orchestration, consistency maintenance, and market distribution ability. Focus shifts from "making video" to "building content systems."

Creation division of labor: Creators/producers define worldview, audience, budget, final judgment; Supervisor Agent interprets director intent, dispatches tasks, maintains project "context"; 90+ specialized agents handle script, storyboard, art, scenes. Humans only do the scarcest "judgment" work; AI handles repetition.

Content form & distribution: Interactive cinema turns viewers from observers to first-person participants. Characters, tasks, scenes react to user behavior; live data feeds next-round plot, version, and distribution decisions. Distribution starts at development: AIGC changes IP discovery, evaluation, localization, monetization — from IP scanning, market signal capture, regional audience prediction, to multi-market version production and licensing operations.

Roundtable: AI Innovation Breakthroughs in Globalization Practice

Bintang Hidayanto (GANI.AI Co-founder): AI value chain = upstream (infra), midstream (apps), downstream (deep business embedding). Upstream globalizes easiest but costs most. Midstream has global potential but requires per-market localization (high effort). Downstream needs high-touch 1:1 service. Companies must know their position. Overhyped: "Agentics" — often misapplied, more fantasy than reality. Underrated: (1) Society still figuring how to turn AI into truly useful tools via continuous experimentation/collaboration; (2) As AI content floods, human creativity (even imperfect) gains value — hence "AI slop" backlash. Market entry: Prioritize high-labor-cost, talent-scarce markets like Japan where AI is a real solution; in low-cost markets like Indonesia, AI is just icing.

Arnd Hungerberg (Pit GTM APAC): Common AI project failures: (1) not mapping/optimizing processes before adding AI; (2) weak data foundations causing hallucinations; (3) no clear success metrics → misaligned expectations; (4) unexpected token cost spikes; (5) employee replacement fears causing resistance. Fix: deeply understand & optimize processes first, set clear AI benchmarks & success criteria, secure top-down and bottom-up buy-in. Overhyped: large-scale human replacement by AI. Underrated: AI's long-term integration into daily life/education — preparing the next generation. Market entry: Start in high-AI-acceptance countries with strong government support: China, Singapore, Korea, Japan.

Mao Hua (2nd Brain CEO, ccMonet.ai founder): By 2030, apps won't vanish but AI Agents will become a new interaction force. Traditional apps hardcode processes/functions in UIs; AI re-integrates scattered capabilities via natural conversation. ccMonet.ai evolves from financial software to AI Finance OS → future AI CFO : 24/7 financial health monitoring, not just error detection but proactive cost optimization, cash-flow risk alerts, new opportunity identification, strategic advice. Overhyped: "AI can do everything" — B2B complexity, data security, privacy, compliance leave huge innovation space. Underrated: "Digitization without dead ends" — vast undigitized niches; AI coding tools (Codex, Claude Code) drastically lower software dev barriers, soon every micro-need could have software. Market entry: Choose Singapore first — mature infra, strong payment willingness, govt tech support, gateway to rest of SEA.