I have spent the better part of my career in financial services — at NM Rothschild in structured finance, in cross-border M&A, in asset-backed securities. I saw trading floors move from open outcry to electronic trading. I watched brokers adopt algorithmic trading systems. I observed risk management transition from rules-based compliance to data-driven statistical models. Now I watch the next transition: from algorithms to AI agents that make decisions with minimal human oversight.
Each transition created new intangible assets — trading models, datasets, risk frameworks, regulatory expertise. Yet financial services companies have been remarkably slow to recognise and value these assets. A bank with a proprietary trading model that generates £200 million in annual profit will record that model on the balance sheet with a value of zero (expensed as an operating cost). A hedge fund with a machine learning model that outperforms the market by 300 basis points will carry that model at zero value in its accounts.
This is changing. Financial services is beginning to recognise AI intangible assets formally — and to structure capital and compensation around them. Understanding what these assets are, why they matter, and how to value them is rapidly becoming a core competency for investors, regulators, and financial institutions themselves.
To understand the intangible assets financial services is creating, it is worth looking at the evolution of the trading floor.
Phase 1 (1980s-1990s): Human traders and institutional knowledge
The asset was the trader — their experience, market intuition, trading relationships, and decision-making heuristics. If a top trader left, the firm lost a valuable asset. Compensation was structured around this: traders were paid massive bonuses specifically to retain the intangible asset of their expertise and relationships.
The intangible assets were the trader's knowledge, market relationships, and reputation. These were impossible to commoditise or transfer. When a star trader left, the knowledge went with them.
Phase 2 (1995-2010): Algorithms and systematic trading
Banks began to codify what traders did — the rules, heuristics, and patterns they followed — into systematic trading models. A team of quants would interview successful traders, extract patterns, translate them into algorithms, and deploy automated trading systems.
The intangible asset now shifted from the individual trader to the algorithm and the team that built and maintained it. A proprietary trading model that generated 20% annual returns was a valuable asset. The model could be deployed across markets, across instruments, across geographies. If a trader left, the algorithm remained.
Banks invested hundreds of millions in building these systems. Yet they expensed the entire investment. The resulting algorithm was worth zero on the balance sheet, despite generating billions in annual profit.
Phase 3 (2010-2020): Machine learning and data-driven models
Machine learning enabled new types of models — ones that learned patterns from data rather than requiring human codification of rules. The investment shifted: instead of teams of quants building algorithms, banks invested in:
The intangible asset became the combination of data and the models trained on that data. A machine learning model for credit risk assessment that used 10 years of proprietary credit data was more valuable than the same model trained on publicly available data.
Financial services firms were now creating genuine data-driven competitive advantages. Yet again, the entire investment was expensed. The resulting models were worth zero on the balance sheet.
Phase 4 (2022-present): AI agents and autonomous systems
The latest phase involves AI agents — systems that can perceive market conditions, form hypotheses about future movements, execute trades, and manage risk with minimal human intervention.
An AI agent for portfolio management might:
The intangible assets underlying these systems are:
Financial services organisations now accumulate intangible assets across five core categories.
A bank's proprietary trading model — the machine learning system that predicts market movements and executes trades — is a genuine, valuable intangible asset.
Cost basis: Training a proprietary trading model might involve:
Total cost to develop a proprietary trading model: £25-45 million upfront, plus annual maintenance.
Income basis: A trading model that generates 200-500 basis points of annual outperformance across a multi-billion pound trading book generates millions in annual alpha.
Valuation: A proprietary trading model generating £60-75 million in annual alpha, with expected useful life of 5-7 years (before competitive models become available or market structure changes), is worth £300-350 million in present value (at 10% discount rate reflecting AI model risk).
Yet this model appears on the balance sheet at zero value — it was expensed as an operating cost.
A financial services firm's proprietary datasets — market data, client data, transactional data — are now recognised as valuable intangible assets under SNA 2025.
What constitutes proprietary data in financial services:
Valuation of proprietary datasets:
Cost approach: A hedge fund with 15 years of accumulative trading data, client data, and alternative data sources might have invested £50-100 million in data acquisition, cleaning, and infrastructure.
Income approach: If the fund generates alpha partly attributable to superior market intelligence from this data, and that alpha is worth £200 million annually, the data asset's share might be valued at £100-150 million (assuming 50-75% of alpha derives from data advantage).
Market approach: Comparable financial data products (Bloomberg, Refinitiv, S&P) sell data subscriptions at £100K-£5 million annually depending on exclusivity and depth. A proprietary dataset that a bank would pay £2-5 million annually to license, over a 10-year horizon, implies a data asset value of £15-40 million.
Practical reality: Most financial services firms carry their proprietary data at zero balance sheet value, despite it being worth tens or hundreds of millions in competitive advantage.
A major bank's risk management infrastructure — the models, systems, and expertise that enable compliance with PRA/FCA regulation and manage counterparty/market/operational risk — is a valuable intangible asset.
This includes:
Valuation perspective:
The risk infrastructure creates value through:
A major bank's integrated risk management infrastructure is easily worth £200-500 million in intangible asset value — yet balance sheets will show it at zero or at a small value for capitalised software.
In financial services, customer relationships are intangible assets with clear economic value.
A customer relationship in banking/wealth management generates:
Valuation:
A bank with 5,000 high-net-worth clients has customer relationship assets worth £25-50 billion. A corporate bank with 200 relationship-managed clients might have customer assets worth £4-20 billion.
Yet these relationships appear on the balance sheet at their carrying value (whatever was paid in acquisitions), not their actual value.
A financial services organisation's ability to navigate regulatory requirements, manage compliance, and maintain productive relationships with regulators is itself an intangible asset.
This includes:
Valuation:
| Asset Category | Valuation Method | Useful Life | Key Risks |
|---|---|---|---|
| Trading models and algorithms | Income (alpha generation) | 3-7 years | Competitive obsolescence, market structure change |
| Proprietary datasets | Income (data-driven advantage) + Market (licensing comps) | 5-10 years | Data commoditisation, new competitor datasets |
| Risk management systems | Cost + Income (capital efficiency, pricing advantage) | 7-10 years | Regulatory changes, model obsolescence |
| Customer relationships | Income (lifetime value) | 10-30 years | Churn, competitive displacement, regulatory change |
| Regulatory relationships/expertise | Income (avoided regulatory costs) | 5-10 years | Staff departure, regulatory turnover, policy change |
From my background in structured finance and asset-backed securities, I see an emerging opportunity: financial services intangible assets are beginning to be structured into capital products.
AI model revenue securitisation: A fintech with proprietary AI models that generate recurring licensing revenue can securitise that revenue in an ABS structure. The cash flows are assigned to an SPV, which issues tranched debt backed by those flows — identical to the lease-backed ABS structures I developed at Rothschild, but backed by software licensing revenue.
Data-backed facilities: A financial services firm with valuable proprietary market intelligence or client data can use that data as collateral for structured lending facilities. The lender takes security over the data (with appropriate safeguards for confidentiality and client privacy) and has a first claim on licensing revenue.
Model-backed equity: An AI-focused fintech can structure its equity to separately value the core AI models and the organisational capital. This enables clearer investment thesis and more efficient capital allocation.
These structures are nascent in financial services, but they are emerging as investors and lenders become more sophisticated about valuing intangible assets.
A hedge fund with a machine learning model that generates £50 million in annual alpha can securitise the expected returns in an ABS structure. Assuming the model has a 7-year expected useful life, the alpha is equivalent to £350 million in present value (at 10% discount rate). An SPV can issue £250 million in senior debt backed by the alpha stream, with the hedge fund retaining £100 million in equity. The fund has effectively converted balance sheet value from an intangible asset that was previously at zero on the books.
Financial services intangible assets — trading models, proprietary datasets, risk management systems — are beginning to be financed through structured capital products. This is expanding access to capital for fintech firms and enabling PE buyers to more precisely value what they are acquiring.
When a financial services firm is sold (to a strategic buyer or to a PE firm), the acquirer will specifically value:
Sellers that have clearly documented these assets — with historical performance data, competitive analysis, and valuation methodologies — will be able to achieve premium valuations. Sellers where these assets remain invisible will be acquiring at a disadvantage.
A bank's AI models and risk systems are only as good as the teams that built and maintain them. The knowledge is partly embedded in the systems (the model weights, the data pipelines), but partly embodied in people (the engineers, quants, traders who understand how to improve and deploy the models).
Financial services will increasingly need to structure compensation and retention strategies around intangible assets. Model-building teams will need to be valued and retained explicitly — not just paid high salaries, but given equity stakes in the models they have built.
Regulators (PRA, FCA, and equivalents globally) are beginning to ask about AI intangible assets in financial services. Questions like:
These questions are precursors to formal regulatory treatment of AI models as capital-relevant assets — potentially requiring additional capital charges for model risk, or explicit disclosure of model performance and governance.
Inventory your intangible assets: What proprietary models do you have? What datasets? What organisational capital and regulatory relationships?
Measure their value: Apply three valuation approaches to each category. Document the historical performance and expected future value.
Document the assets: Ensure models are versioned, performance is tracked, data provenance is clear, and organisational knowledge is documented (not just in people's heads).
Structure your capital accordingly: Make sure compensation, equity structures, and capital allocation reflect the true value of intangible assets.
Prepare for exit: If you expect a potential exit (IPO, acquisition, PE investment), be able to articulate — with data and rigour — what intangible assets the organisation possesses and what value they create.
Financial services organisations have built vast intangible asset bases over the past 20 years — trading models, datasets, risk systems, customer relationships, regulatory expertise. Yet most are valued as if these assets do not exist. The organisations that will extract the most value — at exit, in capital markets, or through new financing structures — will be those that make these intangible assets visible, valued, and managed as the capital assets they truly are.
Tony Hillier is Co-Founder of Opagio. He holds an MA from Balliol College, Oxford and an MBA with distinction. His career includes executive board roles at NM Rothschild & Sons and GEC Finance, and a non-executive directorship at Financial Security Assurance in New York, where he specialised in structured finance, asset-backed securities, and proprietary trading models.
Board oversight of AI investment is becoming as critical as financial controls. Here is the 10-point accountability checklist that separates effective board governance from box-ticking.
Read more →Get the latest insights on intangible asset growth and productivity delivered to your inbox.
Book a free consultation to see how the Opagio Growth Platform can help your business.