NextFin News - Singapore’s latest push to make artificial intelligence more usable inside finance is about more than software adoption. It is a bid to defend one of the city-state’s core economic franchises at a moment when Hong Kong is also pairing finance with faster AI access, broader sandboxes and direct compute support. The question is not whether banks and asset managers will use AI. It is which financial centre will let their best people deploy it quickly enough to stay, and whether that decision will shape where the next generation of high-value finance jobs gets built.
That is a serious policy question because finance is not a side industry in Singapore. According to the Monetary Authority of Singapore, the sector contributes about 14% of gross domestic product, employs around 200,000 people, and added 3,000 net jobs last year. It also sits on top of a capital base that gives the policy debate real weight: Singapore managed S$6.7 trillion in assets as of the end of 2025, while assets in its banking sector grew about 6% annually from 2021 through 2025. This is not a conversation about nurturing a fashionable technology niche. It is about protecting a large, high-margin ecosystem whose value depends on remaining a place where revenue-generating teams, senior control functions and technical specialists want to build careers.
The competitive pressure is also no longer one-dimensional. For years, Singapore and Hong Kong competed on familiar terrain: taxes, regulation, listing activity, travel connectivity, China access, wealth-management depth and legal certainty. Those variables still matter, and they still explain much of the rivalry between the two cities. But the newest layer of competition is whether senior bankers, portfolio managers, quants, compliance executives and private-wealth teams can get access to the tools they increasingly see as productivity multipliers: generative AI applications, internal model infrastructure, validated use cases, reusable workflow components, governance frameworks and the computing power needed to test them in regulated environments.
Singapore has been building that stack steadily rather than theatrically. MAS says more than 30 financial institutions have established AI competency centres in Singapore. It has also expanded PathFin.ai, a knowledge-sharing programme for AI implementation in finance, to more than 200 participating financial institutions. Alongside that, MAS has pushed BuildFin.ai as a collaborative platform for AI-forward financial institutions and, in November 2025, issued proposed Guidelines on Artificial Intelligence Risk Management that apply across the financial sector, covering generative AI as well as emerging AI agents. The strategic message is plain: Singapore wants financial institutions to adopt AI, but to do so through a supervisory architecture that makes scale acceptable rather than reckless.
Hong Kong is pursuing a similarly explicit strategy, and that is why the talent angle matters. On 5 March 2026, Hong Kong’s monetary and market regulators launched GenA.I. Sandbox++, expanding an earlier 2024 initiative across banking, securities and capital markets, asset and wealth management, insurance, mandatory provident fund activities and stored-value facilities. The package goes beyond regulatory signalling. Participating firms are offered targeted supervisory guidance, technical support and complimentary access to graphics processing unit computing resources at Cyberport’s AI Supercomputing Centre. In talent terms, that matters because access friction is often what determines where innovation teams actually sit. A firm may have legal entities in several cities, but the city where a team can test an AI workflow fastest often becomes the city where the next hiring round happens.
The first-order story is straightforward: AI can raise productivity, automate repetitive work, strengthen anti-fraud systems, improve customer response times and reduce the manual cost of compliance. The second-order story is more important. Once firms believe AI-enabled workflows materially raise the output of a portfolio manager, private banker, analyst, risk officer or quant developer, top performers begin to compare jurisdictions not only by compensation and lifestyle but by institutional permissions. The city that offers clearer rules, faster experimentation and easier compute access can become the city where ambitious teams cluster. In that sense, AI access is becoming a competitiveness input in the same way data connectivity, legal certainty and exchange infrastructure were in earlier eras of financial-centre competition.
The Real Battle Is Not for Software but for Permissioned Productivity
The clearest way to understand Singapore’s strategy is to stop thinking of AI as a narrow technology policy and start thinking of it as permissioned productivity inside regulated finance. Large language models on their own are not the scarce asset. The scarce asset is the ability to put those models to work inside institutions that have to answer to regulators, clients, boards and risk committees. In that environment, the winning jurisdiction is not necessarily the one with the loudest consumer AI narrative. It is the one that reduces the institutional friction between curiosity and deployment.
Singapore’s approach is strongest precisely where that friction tends to be highest. MAS is not only encouraging experimentation; it is trying to industrialise trust around AI adoption. PathFin.ai is designed to reduce duplication by letting financial institutions share implementation experience and validated use cases. BuildFin.ai aims to help AI-forward institutions develop reusable components more rapidly and at lower cost. The proposed AI risk-management guidelines attempt to give boards and management teams a common supervisory frame for oversight, controls and life-cycle governance. Taken together, those layers do something important for talent retention: they tell an ambitious team inside a bank or asset manager that Singapore is not merely tolerating AI experimentation but trying to make it operationally legible.
That matters because the people most likely to move between Singapore and Hong Kong are rarely choosing between job and no job. They are choosing between ecosystems. A senior wealth manager deciding where to build a regional book, a hedge fund technologist choosing where to base a platform build-out, or a product executive trying to roll out AI-assisted compliance tools does not just ask where salaries are highest. They ask where projects clear internal approval fastest, where data and model governance are clearest, and where managers believe regulators will support carefully controlled testing rather than treat it as a future problem. In that decision chain, AI access acts less like a perk than like a speed premium.
Hong Kong understands that too, which is why its March 2026 sandbox expansion deserves attention. Complimentary GPU access is not a cosmetic add-on. It attacks one of the practical bottlenecks that can delay AI experimentation, especially for institutions that want to prototype multiple use cases at once. If a team can get supervisory feedback, technical support and compute in the same channel, the innovation loop shortens. That does not guarantee Hong Kong will win the talent contest. But it changes the terms of the contest. The rivalry is shifting from who offers the better headline pitch to who offers the shorter path from pilot to production.
"The launch of the GenA.I. Sandbox++ marks a significant milestone under our ‘Fintech 2030’ strategy, reinforcing our commitment to building a vibrant ecosystem for responsible innovation. By bringing together regulators, financial institutions, and the tech community, we aim to unlock A.I.’s full potential to drive growth, efficiency, and customer-centricity across financial services, further strengthening Hong Kong’s competitiveness as a leading international financial centre." — Eddie Yue, Chief Executive of the Hong Kong Monetary Authority, 5 March 2026
Eddie Yue’s phrasing matters because it makes the competitive objective explicit. Hong Kong is not presenting AI as a side experiment. It is tying AI infrastructure, regulatory support and technical enablement directly to its claim on international financial-centre status. That is why the issue cannot be reduced to one city being more pro-innovation than the other. Both are pro-adoption. What matters is which city is building the more compelling package for institutions whose best staff increasingly expect AI capability to be part of the operating environment.
This is where the story becomes structural rather than merely cyclical. Cyclical competition between Singapore and Hong Kong rises and falls with market volumes, equity issuance windows, mainland deal flow, wealth inflows and macro sentiment. When those drivers turn, activity can rotate back. But AI capability inside finance is not a cyclical sweetener layered on top of the old model. It is becoming embedded infrastructure. Once banks redesign compliance workflows, portfolio analytics, customer onboarding, fraud controls and internal knowledge systems around AI-assisted processes, they do not simply revert to the old model when conditions change. The city that hosts those redesigns gains sticky advantages: management attention, specialist hiring, vendor relationships, training pipelines and eventually local clusters of expertise. That is why the present push looks less like a temporary retention campaign and more like a contest to host the next operating system of finance.
There is still a cyclical leg to the story, and it should be separated rather than blurred. The cyclical leg is the intensity of the contest itself. When capital markets are active, wealth creation is rising and firms are hiring aggressively, the fight for senior producers becomes more visible, compensation becomes more elastic and relocations become easier to justify. When markets slow, that pressure often cools. But the structural leg is the basis on which the contest is now being fought. The rivalry no longer rests only on where clients are, where tax rates sit or where listing volumes are highest. It increasingly rests on whether a financial institution believes a city lets it build faster, with lower internal friction, in an AI-shaped operating model. That part does not self-correct once it is built.
Why Talent Follows Workflow Design More Than Public Slogans
The instinctive read of this story is that top finance professionals move toward money, not technology. There is truth in that. But in modern financial services, technology is increasingly one of the ways money is captured, served and retained. That means talent decisions are being shaped not only by where client assets sit, but by where workflows are redesigned to manage those assets more effectively.
Take private wealth and family-office work. The old model depended heavily on relationship management, product distribution and access to booking centres. The new model still needs those things, but it also depends on how efficiently an institution can personalise reporting, monitor concentrated exposures, automate suitability checks, flag suspicious transactions and surface cross-portfolio analytics. Those capabilities sit at the intersection of finance and AI governance. A private banker or wealth-platform executive may not move because a city has a chatbot pilot. But they may choose to build a team in a city where product, compliance and technology can work off a shared AI control framework without months of delay.
The same logic holds in asset management and hedge funds. A portfolio manager does not need every research process automated to care about AI access. It is enough if AI shortens document review, improves internal search, accelerates coding support, strengthens risk summarisation or reduces the time it takes to transform data into an investment note. Each individual gain may sound incremental. Combined, those gains alter the output of a team. And once firms see that AI-assisted teams can do more with the same headcount, they become more sensitive to where those teams should sit.
This is the deeper transmission mechanism. AI access changes workflow design. Workflow design changes output per employee. Output per employee changes where firms want to locate the next marginal hire, desk or mandate. And once those marginal decisions repeat often enough, they shape the strategic gravity of a financial centre. The visible relocation comes late. The quiet build-out comes first.
That is also why public signalling matters less than implementation density. Singapore can claim to support AI in finance, but the decisive test is whether institutions feel that support in daily operating decisions. Do legal, risk and compliance teams know how to clear an AI use case? Are there shared templates, governance expectations and reusable components that reduce the cost of starting? Are training programmes broad enough that non-technical managers can supervise AI use rather than stall it? A city that can answer yes to those questions earns credibility with the most mobile professionals because it reduces the hidden tax of institutional delay.
Singapore’s official materials suggest it is trying to solve exactly that problem. PathFin.ai is not pitched as a flashy showcase. It is pitched as an industry knowledge hub for AI implementations. BuildFin.ai is not framed as a generic innovation slogan. It is framed as a collaborative platform to facilitate the development of AI applications more rapidly and cost-effectively. The proposed risk-management guidelines are not consumer-facing. They are internal plumbing for boards, controls and oversight functions. In aggregate, the message is that Singapore is trying to make AI deployment governable at scale. That is a more durable objective than simply sponsoring experiments.
But Hong Kong’s offer reveals the risk of a governance-first approach. If one city becomes associated with safe but slow deployment while the other becomes associated with supported and visible experimentation, talent may treat the faster city as the place where careers compound more quickly. That reputational risk is why GPU access, sandbox breadth and cross-regulator coordination matter even if they sound like technical footnotes. They compress time. In talent markets, time is often more valuable than rhetoric.
The Strongest Counter-Thesis Is That Hong Kong’s Asset Gravity Still Beats Singapore’s AI Stack
The strongest case against the Singapore thesis is not that AI is unimportant. It is that AI is still secondary to asset gravity. On that reading, top finance talent will continue to follow the deepest client pools, the most active capital channels and the jurisdictions with the strongest wealth intermediation momentum, even if a rival city offers better AI governance or broader institutional support.
Hong Kong has official data points that make that case credible. InvestHK said the number of single-family offices in Hong Kong surpassed 3,380 by the end of 2025, up more than 25% in two years. Another official release described Hong Kong as Asia’s largest cross-border wealth-management hub and the second largest globally after Switzerland, and said the city is anticipated to surpass Switzerland by 2027. Those are promotional claims and should be read as such, but they still reveal how Hong Kong is positioning itself: as the location where assets, clients and wealth-related opportunity are deep enough to outweigh softer ecosystem considerations.
That argument also fits a long history of financial-centre competition. Senior bankers and portfolio managers often tolerate imperfect infrastructure if the revenue opportunity is large enough. A rainmaker can work around friction. A top asset-gatherer may care more about proximity to clients than about whether internal AI tooling clears one month sooner. In that world, AI access matters at the margin, but not enough to drive the biggest location decisions.
That counter-thesis deserves weight because it attacks the core claim rather than an edge detail. If it is right, Singapore’s AI stack is useful but not decisive. It helps incumbents work better, but it does not change the magnetic pull of a rival hub that offers stronger wealth momentum or superior access to capital formation. That would make AI a retention aid, not a strategic moat.
Still, the counter-thesis underestimates how much finance is changing from a pure relationship-and-balance-sheet business into a workflow-and-control business. The people institutions most want to keep are increasingly hybrid operators: revenue producers who expect analytics leverage, product heads who need automation, and control leaders who now determine how quickly new businesses scale. In other words, client assets and AI capability are no longer separate categories. They interact. A city can attract assets, but if its institutions cannot roll out AI-enabled servicing, risk and personalisation fast enough, the productivity of those assets falls relative to a rival hub. Over time that productivity gap affects where incremental teams are built, and those incremental decisions compound into strategic share.
The real test is not whether AI trumps asset gravity in one leap. It almost certainly does not. The real test is whether AI capability changes the productivity of the people who capture and manage those assets. If it does, then policy around AI becomes policy around talent density, which in turn becomes policy around market share.
The Risk for Singapore Is Not Losing Today’s Jobs but Losing Tomorrow’s Marginal Build-Outs
The most important implication is that Singapore does not need a sudden exodus to have a problem. Structural losses in financial-centre competition usually begin at the margin. The city still keeps most incumbents, but the next trading pod, the next private-bank technology team, the next multi-asset research platform or the next regional AI product lead gets based elsewhere. These marginal decisions matter because they shape the future map of influence long before they show up in headline employment numbers.
That is why the usual debate about whether AI will destroy or create finance jobs is too blunt for this story. The sharper question is where the higher-productivity versions of those jobs will be located. MAS’s own numbers show the stakes. A sector that contributes about 14% of GDP and employs around 200,000 people does not need to lose its current base to feel pressure. It only needs the next wave of higher-value functions to be created elsewhere. In finance, those functions are increasingly tied to data infrastructure, model operations, compliance technology, client analytics, fraud systems and automation-enhanced distribution.
Viewed that way, Singapore’s current AI push looks rational. It is trying to ensure that regulated financial institutions see the city as a place where AI capability can be built with enough confidence, enough scale and enough trained labour to support future business lines. MAS has also made the workforce point directly.
"We want Singapore to be a global launchpad for AI in financial services: to anchor strong AI capabilities within financial institutions here, and to broaden adoption across the sector." — Gan Kim Yong, Deputy Prime Minister and Chairman of the Monetary Authority of Singapore, 25 June 2026
The rest of Gan’s official remarks are equally important. He said more than 30 financial institutions have established AI competency centres in Singapore, that PathFin.ai now has more than 200 participating financial institutions, and that the Institute of Banking and Finance has worked with the Association of Banks in Singapore and the National University of Singapore’s Asian Institute of Digital Finance to launch training programmes on responsible AI use for the financial sector. That combination of adoption, shared infrastructure and workforce preparation is what a structural response looks like. It is an attempt to make the financial sector more productive without severing trust, oversight or local skills formation.
The strongest falsifying signal for the structural thesis would also be clear. If over the next one to two years Singapore’s official AI-in-finance participation metrics continue to rise, but high-value finance build-outs still concentrate disproportionately in Hong Kong, then AI ecosystem development on its own is not the decisive retention lever. A second falsifier would be if firms use Singapore mainly for governance and training while locating revenue-generating AI deployment teams elsewhere. That would suggest Singapore has built the policy wrapper without securing the productive core.
For now, the base case is more balanced. Singapore appears well placed in near-term credibility because it already has meaningful institutional adoption, more than 30 AI competency centres and more than 200 institutions in PathFin.ai. Hong Kong, however, has shown that it is willing to combine cross-regulator coordination with direct GPU support and an explicit competitiveness pitch. The likely outcome is not winner-takes-all but a more segmented rivalry: Singapore leaning into trusted, scalable AI adoption across large financial institutions, and Hong Kong pressing its case as a fast-moving, compute-enabled hub tied to wealth and capital-market activity.
That sets up a three-part outlook. In the short term, the scorecard is about visible implementation velocity: which city can point to more live use cases, faster pilot approvals and clearer evidence that AI is reducing friction inside banks and asset managers. In the medium term, the scorecard shifts to organisational depth: where new teams are based, how quickly AI-driven products clear controls, and whether institutions expand local AI functions beyond proof-of-concept work. In the long term, the decisive variable is whether AI becomes embedded enough in the operating model of finance that hub competition shifts permanently from incentives and geography toward execution capacity inside regulated systems.
The scenario map follows from that timeline. The base case is that Singapore remains competitive because its governance-heavy model proves attractive to institutions that want enterprise-grade deployment rather than narrow pilots, while Hong Kong remains competitive because direct compute access and wealth momentum keep it attractive for faster-moving teams. The upside case for Singapore is that its trust architecture does not merely reduce risk but materially raises deployment velocity, turning governance into a productivity asset rather than a drag. The downside case is that governance depth slows visible speed, allowing rivals to capture the narrative of momentum and attract the next marginal build-outs first.
As of the latest official disclosures used in this article, including MAS materials through 25 June 2026 and Hong Kong official releases through 5 March 2026 and 18 March 2026, there is enough evidence to say the competitive frame has changed even if the final winner has not. The new contest is no longer just about where capital is booked. It is about where finance can be rebuilt around AI with the least friction and the highest trust.
That is the real test. In the next phase of Asian financial-centre rivalry, AI access is not just a technology benefit. It is a way of deciding where the most productive version of finance work gets done.
Explore more exclusive insights at nextfin.ai.
