This article was written by: Tony Livesey, Chief Technology & Product Officer, AutoRek
A managing director walked me through his firm’s data strategy a few months ago, and it was genuinely good. Proprietary flows no competitor could replicate, a plan to turn them into sharper pricing and faster client decisions, the whole case for data as the firm’s defensible edge. Then I asked how they reconcile that data across asset classes today. The answer, after a pause, was a spreadsheet and an analyst who knew which tabs not to touch.
The industry has decided that proprietary data is the competitive moat, the structural advantage that holds off disruption the way scale held off challengers a generation ago. I think that is half right in a way that is more dangerous than being wrong. Data is only a moat if you can control it: reconcile it, validate it, explain it and stand behind it faster than a competitor. A proprietary data estate you cannot control is a cost center with good intentions.
A moat is whether a rival can take your position anyway. Most firms operate with broadly the same categories of data, so ownership decides very little. What decides the contest is trust, the gap between having data and being able to act on it with confidence. And on that test most firms fail, because the advantage never survives the trip from the data estate to the point of decision intact and on time.
AutoRek’s 2026 Investments and Capital Markets Survey tells this story. Across 250 senior operations, finance and technology leaders in the UK and US, 85% said their operational processes already struggle, or will struggle, to scale as volumes grow. Only 14% believed their processes were easily scalable. These are not small or unsophisticated firms. The average respondent processes close to 460,000 transactions a day. They have the data, but simply lack the structural mechanisms required to protect that edge.
Look at where control breaks down. In the same survey, 41% named data integration and compatibility as their single greatest operational challenge, and 82% said a large share of their operational work is still manual. This is the moat failing at its foundation. If your proprietary data has to be pulled from incompatible systems and reconciled by hand before anyone can trust it, then your competitor with cleaner control reaches a decision before you do, using data that is objectively less valuable than yours. They win with a worse asset because they can stand behind theirs and you cannot.
I anticipate some pushback on this point, but I will be direct: for most firms, the investment in AI, in a bid to secure a competitive advantage, may actually be giving their rivals an edge instead.
Adoption is near universal: 98% of surveyed firms use AI in some capacity. Real integration is rare: only 14% have it working across operations rather than in isolated pilots. So the typical firm is pointing intelligent tools at a data estate it already cannot trust.
AI does not fix broken data. It magnifies it. A model run over a fragmented estate does not repair the fragmentation; it produces faster, more confident, more expensive versions of the same errors, at a volume no human reviewer can catch. Used well, AI is a genuine accelerant, but it accelerates control, it does not replace it. In regulated markets credibility matters more than ambition, and the firms that gain from AI are the ones whose control foundation is already sound. For everyone else it widens the rival’s advantage while compounding their own errors, because it speeds up a process that was leaking to begin with.
The reasonable objection is that I am describing one thing as two. A serious data moat already includes the control needed to run it, so why separate the asset from the ability to trust it? In the best firms there is no separation. But in most, the two are split by org chart and by budget. The data strategy is owned in the executive committee and written in the language of differentiation and defensibility. The control layer that would make it real is owned three floors down and funded as a cost to be trimmed. That split is the whole problem. It is why firms can sincerely believe they have a moat while the thing that would make it defensible quietly loses its budget every year. The survey caught the price of that neglect: respondents reported that an average of nearly 16% of operational budgets goes to fixing problems caused by manual processes.
And the demands on internal infrastructure increase every year. 59% of firms now handle digital assets, and 39% call them their single greatest data and operational challenge. Digital assets do not just add volume, they add variability, and variability is what exposes the limits of legacy infrastructure fastest. Every new asset class and data format is another place control can fragment, which is why validating data once overnight no longer holds. The firms pulling ahead are moving toward continuous assurance, applying control as data flows rather than chasing yesterday’s breaks the morning after.
So my prediction, and I will make it specific because a vague one is useless. By 2030 the wealth and asset managers that have actually pulled ahead will not be the ones with the most proprietary data or the best-articulated strategy for it. Everyone will have data and AI. The defensible firms will be the ones that understood the moat was never the data. It was the control to turn data into a decision they could trust, explain and defend faster than anyone could copy, and they funded that control like the competitive weapon it is.
If you want to know whether your firm’s moat is real, do not reread the strategy. Go and watch what your operations can actually do on an ordinary Tuesday, with the systems they have, under the volume they carry. The distance between what your data strategy promises and what your back office can stand behind is the true width of your moat. For most firms, it is narrower than they think, and a competitor with better control is already pulling ahead.