AI skills are now the hardest to find on Earth — and Asia-Pacific has the widest gap of any region. Inside the build-buy-borrow playbook that Singapore, Japan and India are writing for the rest of the world.
Something changed in the global talent data this year. For the first time since ManpowerGroup began surveying employers, artificial intelligence skills have overtaken engineering and traditional IT as the hardest capabilities to find anywhere in the world. Across 39,000 employers in 41 countries, 72% report difficulty filling roles — and nowhere does that difficulty bite harder than in Asia-Pacific, where Japan (84%) and India (82%) report the most severe constraints of any major markets, and where the ratio of demand to qualified supply is the worst on the planet.
Singapore, the region’s talent bellwether, tells the sharpest version of the story. In General Assembly’s State of Tech Talent 2026 — the first edition with a Singapore-specific breakdown — 95% of employers said they struggle to fill tech roles. Not because there are too few people. Because there are too few of the right skills. This article unpacks what is actually happening in APAC’s AI labour market, what governments and employers are doing about it, and what it means for companies and candidates far beyond the region.
A shortage of skills, not headcount
The nuance matters. Singapore’s overall tech hiring difficulty actually eased year on year, from 83% to 71% in ManpowerGroup’s data, as the broader labour market loosened. Yet the shortage did not resolve — it moved. For the first time, AI Model & Application Development (cited by 26% of Singapore employers) and AI Literacy (25%) top the list of hardest-to-find skills, while IT & Data, first a year ago, has fallen to seventh. In General Assembly’s survey, 58% of employers named data analytics and data science roles the most difficult to fill — precisely the capabilities that AI adoption depends on.
In other words, APAC does not have a hiring problem that a bigger pipeline of generalist engineers can solve. It has a readiness problem: the gap between what the existing and available workforce can do and what AI-era roles require. That is why the shortage persists even as unemployment ticks up and the wider talent pool expands — and why employers across the region are quietly concluding that recruitment alone cannot close it.
Why the crunch is worst in Asia
Three forces compound in APAC. First, demand is accelerating fastest there: the region’s AI-in-HR and enterprise-AI markets are growing at the highest rates globally, driven by tech-forward adoption in India, China, Japan and Southeast Asia. Second, the frontier AI industry itself has arrived physically. Thirty of the world’s 250 most valuable AI companies are actively hiring in Singapore as of mid-2026, with OpenAI alone advertising dozens of roles — frontier labs building the commercial teams to sell and deploy AI across Asia-Pacific, and competing for the same scarce specialists as every bank, telco and manufacturer in the region. Third, supply cannot keep pace: estimates put APAC’s shortage ratio at roughly one qualified candidate for every 3.6 openings in key AI roles, the widest gap of any region, with AI compensation in competitive Asian markets inflating 15–20% a year as a result.
Japan illustrates the demographic edge of the problem: the most severe reported shortage of almost any country surveyed (84%), an ageing workforce, and intense demand for technical and R&D talent — which is steadily normalising cross-border teams, remote arrangements and foreign hiring in a market that historically resisted all three. India, despite producing engineering graduates at enormous scale, reports an 82% shortage because demand for genuinely experienced AI practitioners has outrun even its supply. China, notably, reports the lowest difficulty globally at 48% — a reminder that a decade of sustained state investment in AI education changes the curve.
What employers are actually doing: build, buy, borrow
Build: the upskilling turn
The most consequential shift in 2026 is philosophical: 83% of tech recruiters globally now believe company success depends more on upskilling existing employees than on hiring new ones, and the share of companies prioritising internal training jumped from 28% to 35% in a single year. In Singapore, 69% of employers say upskilling will meaningfully reduce their shortages — but 58% cite cost as the primary barrier, a higher share than in the US or UK.
Governments are moving to remove that barrier, and Singapore’s Budget 2026 is the template others will copy: a 400% tax deduction on qualifying AI spending (capped at S$50,000 per company per year), a SkillsFuture Enterprise Credit of up to S$10,000 per employer, and training subsidies covering 70% of AI course costs — rising to 90% for smaller businesses. Institutionally, IMDA’s TechSkills Accelerator is targeting 40,000 tech professionals for upskilling over three years, and a separate national initiative aims to make 100,000 non-technical workers “AI bilingual” by 2029. The policy message is explicit: AI literacy is being treated as workforce infrastructure, not a perk.
Buy: hiring for potential, not pedigree
Where companies do hire, the criteria are changing. Degree requirements and brand-name employers are giving way to demonstrated skills — coding tests, case studies, shipped work — partly because skills-based screening enlarges the pool, and partly because in AI specifically, the most current capabilities are too new for any degree to certify. Employers are also, in ManpowerGroup’s phrase, learning to recruit “AI-enhanced candidates”: professionals whose fluency with AI tools multiplies their output, assessed on judgement and adaptability as much as on syntax.
Borrow: outsourcing and cross-border talent
The third response is the one with the largest cross-border implications: nearly 74% of Singapore companies are already outsourcing tech talent or planning to — in significant part to sidestep visa complications when hiring internationally. Flexible engagement is becoming the default architecture of APAC tech teams: core leadership and product functions concentrated in hubs like Singapore, with engineering and data capacity engaged on contract across India, Vietnam, the Philippines and Eastern Europe. Cost reinforces the logic — comparable AI engineering talent can differ in cost by a factor of two to three between the US and Southeast Asia — but in 2026 the binding constraint is availability, not price: companies borrow talent across borders because they cannot buy it locally at any reasonable speed.
Done casually, this model creates real risk — contractor misclassification, permanent-establishment exposure, payroll non-compliance — which is why the shift has driven demand for agency-managed contracting with compliant in-country payroll rather than informal arrangements.
What it means beyond APAC
For European employers, two implications stand out. The first is competitive: the same frontier labs and well-funded Asian enterprises now bidding up AI salaries in Singapore recruit globally, including from European teams — retention plans should assume it. The second is opportunity: the borrow model runs in both directions, and European specialists in LLM engineering, MLOps and data platforms are increasingly engaged remotely or on contract by APAC organisations that cannot fill roles locally, often at rates reflecting the region’s scarcity premium.
For candidates, the guidance writes itself. The single most valuable career investment in this market is demonstrable AI capability layered onto an existing specialism — the profile every APAC employer names as hardest to find. Portfolios and shipped work now outweigh credentials; regional mobility (or the willingness to work across time zones) commands a premium; and the markets with the deepest shortages, Japan above all, are more open to international talent than they have ever been.
The bottom line
APAC’s AI skills crunch is not a temporary imbalance that graduation cycles will fix; the global shortfall in AI roles is projected to persist into the 2030s. The region’s answer — upskill aggressively, hire for skills over credentials, and borrow talent flexibly across borders — is becoming the playbook everywhere. The organisations that thrive will be those that treat talent as a build-buy-borrow portfolio rather than a requisition list; the candidates who thrive will be those whose AI skills are real, current and provable.
Acquism recruits AI, ML and data talent across Asia and Europe, and manages compliant contract engagement and worldwide payroll for companies building cross-border teams. If you are scaling an AI capability in APAC — or competing with those who are — talk to our consultants about permanent, contract and RPO solutions.