WASHINGTON, October 20, 2026: The International Monetary Fund says financial markets remain resilient but face risks from AI investment, government borrowing costs and cyber threats, while the World Bank examines the technology’s potential to raise productivity and the constraints that could limit its benefits in developing economies.
The institutions address different aspects of the changing global outlook. The IMF examines vulnerabilities that could emerge if financial conditions deteriorate or returns on AI investment disappoint. The World Bank’s regional assessments consider how technology production, exports and adoption may affect growth and employment.
The implications vary across economies. Countries producing AI-related goods are benefiting from investment demand, while others face high energy costs, expensive financing and gaps in infrastructure and skills. The scale and distribution of future productivity gains remain uncertain.
Global growth holds as financial risks persist
The IMF’s October World Economic Outlook projects global economic growth of 3.1 per cent in 2026 and 3.4 per cent in 2027. Global headline inflation is projected to reach 4.7 per cent in 2026 before easing to 4.1 per cent in 2027.
Investment associated with artificial intelligence is supporting economic activity in some markets, but the outlook remains exposed to conflict, energy prices, fiscal vulnerabilities and renewed trade tensions. The IMF identifies a reassessment of expectations surrounding AI investment as one of the risks that could weaken growth.
The World Bank’s January Global Economic Prospects analysed how a financial-market correction could affect the outlook. Its downside scenario estimated that falling share prices, weaker investor confidence and tighter financial conditions could reduce global growth by as much as 0.3 percentage points against its 2026 forecast. Stronger productivity gains from AI could, in the opposite direction, improve the outlook.
The Peterson Institute for International Economics offered a separate assessment in its global growth analysis. It projected growth of 3.2 per cent in 2026 and 3.1 per cent in 2027, with AI investment supporting activity even as energy prices, inflation and borrowing costs weigh on the economy. These projections come from a different institution and forecasting framework and should not be treated as directly interchangeable with the IMF’s estimates.
AI production is reshaping regional growth
The World Bank’s East Asia and Pacific Economic Update projects regional growth of 4.5 per cent in 2026. Vietnam is forecast to grow by 7.4 per cent, Malaysia by 5.1 per cent and Thailand by 2 per cent. The Bank attributes stronger activity in several economies partly to the manufacture and export of high-technology goods amid the global surge in AI-related demand.
Participation in technology supply chains is an important factor in the regional outlook. Economies producing electronics and other technology-related goods can benefit from rising demand for computing infrastructure, although this does not necessarily mean that AI adoption is equally advanced across their domestic businesses. High energy costs and weaker global demand remain risks.
The outlook is more difficult in the Middle East, North Africa, Afghanistan and Pakistan. The World Bank’s regional economic update projects the regional economy will contract by 2.1 per cent in 2026, following growth of 3.3 per cent in 2025. The Bank attributes the deterioration to conflict and its economic consequences, including disruption to energy production, transport, tourism and trade.
The same report examines AI as a potential source of longer-term productivity growth. It estimates that AI could improve productivity in 13 to 20 per cent of jobs across the region, while fewer than 10 per cent face near-term automation risk. These estimates describe potential effects, not productivity gains already achieved. Infrastructure, skills, local-language data and the capacity of businesses and public institutions to use AI will influence the results.
Gulf governments and technology companies are investing in digital infrastructure and public services. The United Arab Emirates announced a framework to introduce AI systems across 50 per cent of government sectors and services within two years, according to the UAE government.
Microsoft has announced plans to invest more than $10 billion across the Middle East by 2030, covering capital and operating expenditure associated with cloud and AI infrastructure and services in Kuwait, Qatar, Saudi Arabia and the UAE.
The Middle East Institute examines Gulf governments’ AI ambitions and their links to economic diversification. A separate Carnegie Endowment analysis considers whether data centres could allow energy-producing economies to monetise their resources through digital infrastructure without necessarily reducing their dependence on fossil fuels. Building data centres alone does not establish that a country has developed the research capabilities, skilled workforce or domestic technology industry needed to compete across the AI economy.
Productivity potential meets structural constraints
The World Bank’s World Development Report 2026 evaluates AI’s potential effects on jobs and productivity across economies at different income levels. Its estimates put the share of jobs at risk of automation at 4.5 per cent in low- and middle-income countries, compared with 14.2 per cent in high-income countries. It also estimates that AI could meaningfully improve productivity in 16.2 per cent of jobs in developing economies.
These estimates reflect differences in occupational structures and the tasks current AI systems can perform. However, they do not mention how many jobs will disappear or how much productivity countries will ultimately gain. Adoption, investment, worker training and changes in business practices may influence the outcome.
An IMF working paper, AI and Economic Divergence in Asia, investigates how differences in adoption timing and economic preparedness could affect growth across the region. Its model suggests that advanced economies may adopt AI earlier, while developing economies could face delays and initial growth pressures as capital costs rise. Gains may also differ according to workers’ skills and ownership of productive assets. As an IMF working paper, the analysis does not necessarily represent the views of the Fund or its Executive Board.
AI developer Anthropic offers a separate estimate of potential productivity gains. Its January 2026 Economic Index estimates that AI could add 1.8 percentage points to annual US labour productivity growth over a decade. The estimates fall to 1.2 percentage points when based on Claude.ai usage and 1.0 percentage point when based on API usage. These are model-based estimates derived from patterns of AI use, not measured increases in national productivity.
Labour-market research has yet to establish the full employment effects. A Brookings analysis describes the research as being at an early stage, with limited data to separate AI’s effects from other changes in the economy.
A study of Danish workers published by the National Bureau of Economic Research (NBER) found no measurable average changes in earnings or working hours over the period monitored, although workers reported changes in tasks and new AI-related duties.
Employer surveys describe anticipated changes rather than completed job losses. The World Economic Forum’s Future of Jobs Report 2025 found that 77 per cent of surveyed employers planned to retrain or upskill workers between 2025 and 2030, while 41 per cent expected to reduce their workforce where AI could replicate tasks.
Workers’ organisations have called for a role in decisions about how AI is introduced. The International Transport Workers’ Federation, in its principles for AI in transport, calls for worker participation, consultation, training and safeguards. The International Labour Organization’s tripartite conclusions on AI in manufacturing set out recommendations developed by government, employer and worker representatives. The conclusions were scheduled to be submitted to the ILO Governing Body in November 2026 for consideration.
US research has identified changes among some younger workers without establishing a broad economy-wide employment decline attributable to AI. Stanford Digital Economy Lab’s Canaries in the Coal Mine? examined payroll data through June 2026. It reported that employment among workers aged 22–25 in AI-exposed occupations was 19 per cent below an estimated counterfactual, while cautioning that the entire gap could not be attributed to AI.
A Federal Reserve Bank of New York staff report examined more than 1.6 million workforce-training participation spells from 2012 to 2023. Among participants in occupational training, those in AI-exposed occupations had higher subsequent earnings on average than comparable participants receiving job-search assistance. The analysis concerns training outcomes; it does not establish that all participants had lost jobs because of AI.
The financial consequences of an AI slowdown
The IMF’s October Global Financial Stability Report says financial markets have remained resilient, but financial stability risks remain elevated. A decline in expected AI investment or profitability could lead to a correction in AI-related asset prices. Concentrated exposures and rising leverage could amplify the effects, while sharp increases in sovereign bond yields and cyber incidents pose additional risks.
The scale of planned technology investment is central to this assessment. S&P Global Ratings’ October analysis estimates that capital expenditure by six major cloud providers could rise from $470 billion in 2025 to $870 billion in 2026 and $1.3 trillion in 2027. The ratings agency expects the spending to reduce some providers’ financial flexibility over the following one to two years, while noting that profitable core businesses and predictable revenue support their credit ratings.
Company disclosures describe both demand for computing capacity and the cost of meeting it. Microsoft reported $41 billion in quarterly capital expenditure in its fourth-quarter fiscal 2026 earnings call. About two-thirds went towards short-lived assets, including central processing units and graphics processing units. The company said demand exceeded available capacity and expected capital expenditure to rise in fiscal 2027 while remaining free-cash-flow positive.
Alphabet’s second-quarter 2026 filing with the US Securities and Exchange Commission reported that Google Cloud revenue rose 82 per cent year on year to $24.8 billion. The filing also recorded $49.6 billion in equity issuance in June for general corporate purposes, including AI capital expenditure, and $20.3 billion in net proceeds from senior unsecured notes.
The Bank for International Settlements’ January analysis of AI financing examines the growing role of debt and private credit in financing the expansion. It described immediate risks as moderate but noted that the sustainability of investment depends on future earnings and the relationship between asset valuations and returns.
The corporate disclosures describe investment, revenue and demand for computing capacity. The financial assessments address whether expected returns can support the scale of spending, how borrowing affects companies’ financial flexibility and how a market correction could spread through the financial system.
Developing economies face different paths into AI
The concentration of AI research and infrastructure creates a challenge for countries seeking to participate in the technology economy. UN Trade and Development (UNCTAD), in its Technology and Innovation Report 2025, reported that 100 companies accounted for more than 40 per cent of global business investment in research and development. The report also described the concentration of frontier technology research and development among companies based in developed economies. Separately, it reported that the United States and China accounted for around one-third of peer-reviewed articles and about two-thirds of patents in frontier technologies.
These figures concern different measures of technological capacity: business investment in research and development, scientific publications and patents. They describe the concentration of investment and research, not the distribution of AI use or the productivity gains countries may achieve through adoption.
Some developing economies are seeking to build domestic capacity through public investment. Brazil’s Artificial Intelligence Plan for 2024–2028 set out planned spending of 23 billion reais over four years for innovation, infrastructure, skills, public services and governance, including a high-performance supercomputer. The plan sets out policy commitments and proposed investment; it does not establish that all the funding has been spent or that the intended results have been achieved.
The IMF’s analysis of economic divergence in Asia and the World Bank’s work on AI adoption identify skills, infrastructure, finance and institutional preparedness as factors shaping countries’ ability to benefit. National plans, private investment and access to computing resources represent different routes into the AI economy, with results depending on implementation and the ability to turn investment into productive activity.
What remains uncertain about AI-led growth
The IMF’s financial assessments and the World Bank’s development research address different parts of the changing economic landscape. One examines vulnerabilities that could emerge if valuations, borrowing conditions or investor expectations shift; the other considers how AI might affect production, jobs and productivity across economies.
Planned capital expenditure and rising revenue at technology companies do not establish the long-term returns on the wider AI investment cycle. Estimates of productivity potential likewise do not show how quickly gains will appear or how they will be distributed among firms, workers and countries.
The World Bank’s World Development Report 2026 examines how concentration in advanced AI models, chips and data centres can create dependencies for developing economies, while adaptation to local languages, institutions and development needs can influence how much they benefit. Its analysis adds a further dimension to the IMF’s financial-risk assessments, regional growth projections, company investment disclosures and labour-market research: how far AI investment translates into sustained productivity growth, and which economies and groups capture the resulting gains, remains unsettled.
