Finance Industry’s AI Spending Surge Exposes a Growing Data Divide

Artificial intelligence spending is entering a new phase in the global investment-management industry. Clearwater Analytics’ “GenAI and the Data Divide” study, based on responses from 178 senior executives at asset managers, insurers, hedge funds and private-credit firms across the United States, Europe and Asia, found that 95 per cent of participating organisations had increased their AI budgets over the previous year. More significantly, 85 per cent expected to raise their budgets by at least another 50 per cent during the coming 12 months. The figures demonstrate that AI is no longer being treated as a temporary innovation project. It is becoming part of the long-term operating model of financial institutions, influencing technology investment, recruitment, risk management and the design of investment products.

The expected impact extends across the front, middle and back offices of financial firms. Around 62 per cent of surveyed fund managers believe AI will bring major or transformative changes to the generation and summarisation of financial data. This includes producing investment reports, processing earnings announcements, consolidating portfolio information and converting large volumes of market data into shorter analytical outputs. Another 58 per cent expect substantial changes in decision-support systems, including tools that recommend portfolio rebalancing actions based on defined risk limits and investment objectives. Meanwhile, 57 per cent anticipate that AI will transform predictive modelling and stress testing, allowing firms to analyse more scenarios, respond faster to market movements and evaluate how portfolios may perform under extreme economic conditions.

These applications could substantially improve productivity, but the results depend on the quality of the underlying information. The Clearwater study identified a 23-percentage-point gap between confidence in data completeness and confidence in data accuracy. While 79 per cent of respondents considered their organisations’ data sufficiently complete, only 56 per cent believed it was accurate. A company may therefore possess large volumes of portfolio, transaction and market information without having the consistent classifications, reliable historical records or integrated systems required to train and operate dependable AI models. When information is fragmented across legacy systems, entered under different standards or duplicated between departments, AI can process errors faster without necessarily producing better decisions.

The problem becomes more serious as AI moves from summarising documents to recommending or executing financial actions. An incorrect data field in an ordinary report may be inconvenient, but inaccurate information used in portfolio construction, credit assessment, regulatory reporting or liquidity management could generate financial losses and compliance failures. Firms must also account for model hallucinations, cyberattacks, biased outputs and the limited explainability of complex systems. In a regulated industry, an AI-generated recommendation cannot simply be accepted because it appears plausible. Institutions need clear audit trails, validation procedures, access controls and accountable human decision-makers who can explain why a model was used and how its output affected a financial decision.

Broader industry research suggests that this divide is already affecting financial performance. A global study led by the Cambridge Centre for Alternative Finance found that AI adoption has reached most financial institutions, but advanced implementation and measurable profitability remain uneven. Institutions with greater spending, stronger technological foundations and better-prepared workforces are more likely to report financial benefits, while many organisations have yet to see a material change in profitability. This suggests that purchasing additional models or computing capacity is not enough. Successful implementation requires investment in data engineering, employee training, governance, cybersecurity and organisational redesign alongside the technology itself.

The emerging competitive divide will therefore be between institutions that build reliable AI operating systems and those that accumulate disconnected AI tools. Leading firms are likely to create centralised data foundations that connect investment, accounting, risk and compliance information while applying common standards across the organisation. They will also train portfolio managers and operational employees to understand both the capabilities and limitations of AI. The immediate race may appear to be about spending, but the longer-term advantage will come from trust. Financial institutions that can produce accurate, explainable and well-governed AI outputs will be able to automate more complex functions with confidence, while firms with weak data foundations may discover that larger budgets merely increase the scale of their existing problems.