The rise of data as a financial asset

For most of the modern business era, data has been treated as a by-product: something generated in the course of operations, used to inform decisions, and otherwise left to accumulate. That position is changing.

Data is increasingly recognised, valued and transacted as a financial asset in its own right, with measurable worth, identifiable buyers and a growing role in how businesses are valued and financed.

This shift matters because it exposes a gap. Many organisations hold substantial proprietary data that carries real financial value, yet continue to treat it as worthless simply because it does not appear as an asset in their accounts. Understanding how data is becoming a financial asset and why reported value lags economic value is becoming an important consideration for management teams, investors and acquirers.

From operational resource to recognised asset

The economic status of data has moved.

In its 2025 update, the System of National Accounts reclassified data as a produced capital asset, placing it alongside software, machinery and intellectual property as a productive resource expected to generate future economic benefit. Standard setters have moved in a similar direction, offering clearer guidance on how enterprises should account for data resources rather than leaving them unrecognised. In principle, the long-running question of whether data belongs on the balance sheet has been settled in favour of the view that it does.

Corporate reporting conventions, however, remain some way behind. Traditional accounting was built around physical assets and conventional intellectual property, which have clear ownership, pricing and transferability. Where intangibles are recognised, they are generally assumed to depreciate and are amortised accordingly. In practice, a well-maintained proprietary dataset often does the opposite, becoming more valuable over time as it accumulates history and depth. The result is that reported value frequently bears little relationship to underlying economic value.

The market is already treating data as an asset

The clearest evidence of this shift comes not from accounting theory but from market behaviour.

Data is increasingly packaged and sold as a discrete product, with defined pricing, access tiers and renewal cycles. Providers such as S&P Global offer reference datasets covering more than 1.5 million instruments as standalone product lines, where the data is not a resource supporting the business but the revenue-generating asset itself.

The infrastructure underpinning the data economy is also being drawn into structured finance. Data centres are now routinely securitised, with approximately $11 billion of data centre asset-backed securities issued across 23 deals in 2025 and a further $30 to $40 billion of issuance anticipated across 2026 and 2027. Cash flows associated with storing and processing data are being pooled and sold to investors in the same way as more established asset classes.

This activity relates primarily to the physical layer of the data economy, where valuation is most straightforward, rather than to proprietary datasets themselves. The significance lies in the direction of travel. Once an asset class can be reliably valued and its cash flows modelled, financial markets tend to develop mechanisms to trade, package and lend against it. Proprietary data sits earlier on that trajectory rather than outside it.

Why artificial intelligence has accelerated the shift

Data would not command asset status if it were abundant and interchangeable. Its financial significance has risen sharply because artificial intelligence has made distinctive data scarce in a way that models are not.

As AI systems become more capable, the models themselves are increasingly commoditised. Access to differentiated, proprietary data is not. A competitor can license the same foundation model, but cannot easily reproduce years of accumulated customer behaviour, operational outcomes or longitudinal records. That asymmetry is precisely what gives an asset value: difficult to replicate, but productive once held.

Investors have begun to formalise this into a test. A data advantage, or "data moat", is often defined as one a competitor would need roughly three or more years to replicate, combined with clear evidence that customers will pay for what the data enables. Bloomberg illustrates the point: four decades of proprietary financial data integrated into traders' workflows, giving it a replication timeline measured in years rather than months. Amazon is another, its proprietary commerce data supporting an advertising business that reached roughly $68 billion in revenue in 2025. In each case the value derives not from the volume of data but from the fact that it is uniquely sourced and difficult to reproduce.

Value that already exists but goes unrecognised

One consequence of this shift is that many organisations are holding financial value they have not identified.

Years of ordinary operations leave businesses with large volumes of proprietary information that, because it does not appear as a distinct asset in the accounts, is assumed to have no value beyond its operational use. In many cases that assumption is mistaken. Where data is unique, high in quality and subject to identifiable demand, it can be realised through several established routes:

• Licensing: granting third parties access to proprietary datasets they value but cannot easily obtain, creating a recurring revenue stream.

• Data products: structuring and packaging information for sale as a product in its own right.

• AI applications: supplying differentiated data for model training and validation, where it is increasingly sought after.

• Partnerships and research: using datasets to underpin commercial collaborations or research initiatives.

  
  

The asset, in these cases, already exists. What is often missing is the recognition that it carries value beyond its original purpose, and a route to realise it.

Not all data is a financial asset

Volume alone does not create value. Many organisations hold large quantities of information that are useful internally but have limited standalone worth. Financial value tends to arise only where data has characteristics that are difficult for others to replicate: uniqueness, quality, longitudinal depth, clear ownership and usage rights and identifiable demand. A smaller, highly differentiated dataset can be worth considerably more than a far larger one that lacks these qualities. The relevant question is not how much data an organisation holds, but whether that data can produce a benefit others cannot easily recreate.

Implications for management teams and investors

For management teams, the shift calls for a change in how data is regarded. Rather than treating it as a settled operational resource, organisations should assess what proprietary data they hold, whether it meets the tests of a genuine asset, and which realistic routes to value might apply.

For investors and acquirers, the same logic points to an opportunity in mispricing. As so few businesses carry data at anything close to its economic worth, data value is frequently embedded within a company at a discount, folded into goodwill or overlooked altogether. Distinguishing datasets that represent real, realisable value from those that are simply large is becoming a meaningful source of advantage, particularly in sectors such as healthcare, financial services and technology, where data often forms part of the strategic rationale for a transaction.

The rise of data as a financial asset reflects a genuine change in how value is created and measured. The economic frameworks have begun to recognise it, and financial markets are already transacting around it. For organisations that hold differentiated, high-quality and commercially relevant datasets, data represents an asset with real financial value, whether or not that value currently appears in the accounts.

The opportunity lies in recognising that value, testing it honestly, and finding the routes to realise it, rather than assuming the balance sheet has already told the full story.