AI Comes for the Analysts: Hong Kong and Singapore's Finance Sector Reckons With the Machine

When Goldman Sachs announced in 2025 that its AI systems were handling the work previously done by hundreds of junior analysts across mergers and acquisitions, the announcement landed quietly. The numbers were disclosed in an earnings call. There was no press release, no ceremonial laying-off of analysts on a trading floor. The work simply stopped being assigned to people.

Eighteen months later, that quiet reallocation is rippling outward from the trading floors of Wall Street to the glass towers of Central and Marina Bay. According to reporting by Nikkei Asia published on 29 May 2026, Hong Kong and Singapore — the region's twin pillars of financial services employment — are now directly in the path of what the publication describes as an "AI chill" settling over the graduate finance job market. Asia's financial hubs, it notes, are "on the front lines of an intensifying global battle to find enough jobs for graduates in the age of artificial intelligence."

The story is not that artificial intelligence is arriving in finance. The story is that it is arriving faster, and with greater displacement of entry-level roles, than the institutional infrastructure designed to absorb Asia's finance graduates ever anticipated.

The Analyst Pipeline Runs Dry

The traditional architecture of a finance career has not changed fundamentally in forty years. Graduate enters bank as analyst. Analyst spends two to three years building financial models, preparing pitch decks, and running due diligence spreadsheets under the supervision of an associate. Associate is promoted or departs. Graduate becomes associate. The ladder climbs upward, and with it, the region's supply of junior labor keeps the senior machine running.

That pipeline is now encountering a structural rupture at its base. The spreadsheet work, the model-building, the comparable transaction analysis — tasks that once required a new graduate twelve weeks of training to perform — are being absorbed by large language model systems fine-tuned on decades of deal data. Financial modeling software from vendors including Bloomberg, FactSet, and a clutch of Singapore-based fintech startups has reached a threshold where it can generate a first-draft DCF model from a public filing and a set of analyst prompts faster than a human analyst can open a laptop.

This is not, practitioners are careful to note, a story of mass lay-offs in the conventional sense. The headcount at major banks operating in Hong Kong and Singapore has remained relatively stable in the past eighteen months. What has changed is the entry door. Several international banks with large graduate analyst programmes in Asia told Nikkei Asia that they are reducing cohort sizes for 2027 entry by between 20 and 40 percent. The roles being cut are not the relationship management positions or the deal-origination jobs that require face-to-face client interaction. They are the structural back-office roles — the financial analysis, the data reconciliation, the report generation — that form the traditional on-ramp to a finance career.

The career architecture assumed a division of labor in which humans did the analytical spade work and machines did not compete. That assumption is now obsolete.

The Counter-Argument: Human Judgment Has Not Been Automated

To hear the region's banking executives tell it, the disruption is real but the apocalyptic framing misses something important. The functions being automated are precisely the functions that junior analysts found most alienating — the repetitive number-checking, the deck-polishing, the overnight revision cycles that burned out a generation of analysts without necessarily producing better financial outcomes.

A senior managing director at a US investment bank with significant Hong Kong operations, speaking on condition of anonymity because they were not authorized to discuss staffing strategy, told this publication that the automation of routine analytical tasks is, in their view, a long-overdue correction rather than a crisis. "We are not reducing our need for smart people," the director said. "We are redistributing what we ask them to do. The graduates who can navigate ambiguity, manage a client conversation, think about non-quantitative risk — those people have never been more valuable."

This argument has institutional backing. A 2025 survey of graduate recruiters at financial institutions operating across Asia, cited in the Nikkei Asia reporting, found that while technical financial modeling skills had dropped in stated priority from 67 percent of hiring criteria in 2023 to 41 percent in 2025, competency in client communication, regulatory navigation, and cross-border deal coordination had all risen. The implication is that AI has not eliminated the need for judgment — it has changed which judgments matter most at entry level.

The structural reconfiguration, proponents argue, could ultimately produce more interesting careers for the graduates who make it through. An analyst no longer building DCF models from scratch is instead working directly with a senior managing director on client strategy from their second month, rather than their fifth year. Whether that reconfiguration happens smoothly — or whether the ladder simply removes its bottom rungs — is the contested question.

Two Cities, Two Models, One Disruption

Hong Kong and Singapore have long competed for the same pool of regional financial talent and the same category of global investment flows. Their responses to the AI challenge, however, reveal different institutional instincts.

Singapore has moved faster toward an explicit policy response. The Monetary Authority of Singapore has published guidance on workforce transition in financial services, and the city-state's three major universities have each restructured finance and economics programmes to include mandatory coursework in data literacy, algorithmic tools, and the ethical governance of automated financial decision-making. The government's SkillsFuture framework, which provides subsidized retraining credits to workers at every career stage, has been extended to mid-career financial analysts displaced by automation in the past two years. Singapore's approach treats AI-driven labor market disruption as a policy problem with policy solutions.

Hong Kong's response has been more diffuse. The Hong Kong Monetary Authority has engaged with the technology transition largely through supervisory guidance on model risk management — essentially, rules governing how banks deploy AI in customer-facing and risk-management functions — rather than through direct劳动力 market intervention. The city's universities have expanded fintech offerings, but at a slower institutional pace than Singapore's, partly reflecting a governance culture in which universities operate with greater autonomy from central workforce planning. Hong Kong's financial institutions have, by default, left the adaptation problem largely to the market and to individual workers.

Neither approach is obviously superior, and both face the same underlying constraint: the pace of AI capability development is outrunning the institutional capacity to plan for its labor market effects. A policy framework developed in 2024 looks different in light of what generative AI systems can do in 2026. The disruption is moving faster than the adaptation.

Precedent: The Pattern Looks Familiar, but the Scale Is New

Financial services has absorbed technological disruption before. The introduction of electronic trading in the 1980s and 1990s decimated floor-based trading roles and created entirely new categories of employment — algorithmic trading, electronic market-making, quantitative analytics — that did not exist before. The adoption of Bloomberg terminals and integrated data platforms in the 1990s and 2000s eliminated large numbers of manual data aggregation roles while expanding the demand for professionals who could interpret and act on the newly abundant data.

In each previous cycle, the net employment effect over a ten-year horizon was neutral to positive: roles were eliminated, but new roles emerged faster than they were destroyed, and the overall size of the financial services workforce grew. The transition was painful for individuals, often concentrated among older workers who had built expertise in soon-to-be-automated skills, but the system as a whole absorbed the shock.

The current wave differs from those precedents in two structural ways that are worth naming plainly. First, the previous cycles automated physical tasks — the physical act of executing a trade, the physical aggregation of data from multiple sources — while preserving the analytical and interpretive functions that humans performed. The current wave is automating cognitive tasks that previously required years of training to perform competently: financial modeling, regulatory document interpretation, initial due diligence on target companies. The automation is moving up the value chain, not just completing the elimination of manual labor.

Second, the speed of adoption is faster than in any previous cycle. Large language model systems capable of performing financial analysis tasks at a baseline professional standard became commercially available in 2023. By 2025, they had been integrated into the major financial data platforms used by virtually every institutional investor and bank in Asia. The adoption curve is steeper than anything the industry has experienced since the introduction of electronic trading, and the previous electronic trading transition took nearly a decade to fully propagate through the industry.

Whether this cycle follows the precedent and produces net-positive employment effects over a ten-year horizon, or whether it represents a genuinely discontinuous break with that pattern, is a question the available evidence does not yet settle. The graduates entering the job market in 2026 are, in effect, the first cohort to navigate the new terrain without the benefit of historical analogy.

The Stakes: Who Wins, Who Loses, and Over What Horizon

The most direct losers in the near term are clear. Graduates who built their undergraduate and early-career trajectories around traditional financial analyst roles — the spreadsheet-intensive, model-building, deal-support functions that formed the reliable first step of a finance career — face a job market that is structurally smaller than the one their predecessors entered. The competition for the remaining entry-level roles has intensified commensurately. For a cohort of graduates who selected mathematics, economics, or finance degrees on the assumption that a major bank's analyst programme would be waiting at the end, the arrival of AI is not an abstraction. It is a foreclosed option.

The beneficiaries in the near term are the financial institutions themselves, which are achieving significant cost reductions in back-office and analytical functions. A first-year analyst at a major investment bank in Hong Kong commands a base salary of approximately HK$1.1 million per year, plus bonuses that can double total compensation. Automating the analytical work that justified those salaries, while retaining the client-facing and origination functions, represents a meaningful improvement in the economics of financial intermediation. Those savings accrue to shareholders and, in competitive markets, to clients in the form of lower fees.

Over a longer horizon — five to ten years — the distribution of winners and losers becomes more difficult to trace. If the automation of routine analytical work genuinely frees senior bankers to focus on higher-value client relationships and more complex transactions, the net demand for experienced financial professionals could rise even as entry-level demand falls. That outcome would concentrate the profession's rewards at the top and make career entry more contingent on networks, prior experience, and non-cognitive competencies that are more difficult to develop through formal training.

There is also a regional dimension. Hong Kong and Singapore compete for the same talent pool, the same institutional mandates, and increasingly the same AI-enabled financial infrastructure. If one city-state manages the transition more effectively — producing graduates better adapted to working alongside AI systems, offering better retraining pathways for displaced analysts — it could draw financial talent and institutional investment away from the other. The AI chill is not just a labor market story. It is a story about the relative attractiveness and resilience of Asia's two premier financial centres.

What is clear is that the pipeline that once reliably produced a generation of finance professionals — enter at the bottom, climb to the top — is being redesigned from below. The architects of that pipeline are the systems that can now perform the entry-level tasks that once justified hiring graduates in the first place. The graduates are waiting to see what the new architecture looks like, and who it is designed to serve.

— Monexus has covered Hong Kong's economic trajectory and Singapore's financial sector evolution separately; this article represents the first integrated long-form examination of how AI-driven labor market disruption is testing the competing models of both cities simultaneously.