AI is here…and here …for finance. But it’s not that simple.
Finance has built a powerful mythology around data advantage. But, despite the availability of advanced technology and information delivery methods, much of today’s Wall Street still operates on shockingly thin, fragmented, and poorly managed data. It’s been that way since the 1980s.
Bloomberg terminals, S&P Global feeds, and other vendor platforms actively monetize the scarcity. An industry within finance exists to sell limited datasets as premium intelligence (often at an equally premium price). That likely won’t change, but it’s clear there’s an industry-wide need for significantly more data. And fast.
It’s coming to a head now because generative AI exposes data limitations in finance quicker than any technology before it. There’s no hiding faulty or limited sets under binders in the data guy’s cubicle in the back of a hedge fund’s Tulsa office. LLMs need an insane amount of data points to return context-rich analysis and recommendations that make using the tech worth it.
Gen AI, powered by LLMs,, is a bonafide stress test. And finance is failing.
The main reason: LLMs are intensely unforgiving. When data to justify AI’s use in investment research, risk management, or trading strategy isn’t there—or isn’t usable—the results are shallow, brittle, or misleading. It’s no different from the current model of limited information zhuzhed up to sell as premium, subscription-unlocked data.
I’m not saying hedge funds or asset managers are starving for data. But the information they do have is narrow, inconsistently structured, consistently poorly labeled, and often locked inside legacy systems. Many firms still use platforms adapted from data dashboards that were never designed for real-time inference or large-scale synthesis. Both crucial to unlock ROI from gen AI adoption.
For decades, those shortcomings were manageable. Fund managers, researchers, and analysts operated within spreadsheet cells, in Wall Street boardrooms, and over the phone.
With gen AI in the picture, there is no hiding these limitations behind a “data guy” in a back office anymore. The constraints are visible immediately. AI cannot and will not paper over gaps, apply judgment, and compensate for missing context the way human counterparts have been doing since the 80s.
That’s why we remain in AI for finance purgatory. Firms experiment with copilots that summarize research notes or draft emails, but stop short of deploying AI directly into capital market workflows. Each time, data—or lack thereof—is the bottleneck; not AI. And firms aren’t moving beyond demos or small programs taking place away from capital markets.
However, with the recent boom of vertical-specific AI products following the initial rollout of ChatGPT, there’s more awareness of gen AI’s potential and industries falling behind the technology like finance.
On a brighter note, there’s ample opportunity for adoption and iteration. In fact, I’d say the current data available to hedge funds is 1/100th the amount needed to get the most out of gen AI for finance.
That's a lot of green space to work with.
But progress, like in many industries, requires letting go of the old way of doing things. True AI adoption among hedge funds and the broader financial industry requires a complete overhaul of data infrastructure and systems designed to support it.
The internet is now fundamental to capital markets. Gen AI is next. As an industry, we need to move collectively toward a future where AI augments capital market workflows and individuals, rather than replaces them outright or generates emails, notes, or memos so they can focus on trading.
Duct taping gen AI tooling to old systems, internal folders filled with chaotic spreadsheets, or workflows won’t change anything.
Claiming 100% data accuracy as an AI provider won’t do it either because that’s literally impossible in a forever-changing environment like capital markets.
Firms should lean into the fact that gen AI exposes cracks in the system; not run away from it. Take the opportunity to lay everything bare, modernize data management, and expand data resources to close gaps in context and usability.
Hedge funds and research firms monitoring public companies powering global economic sectors from transportation to tech to industrial materials to healthcare can benefit from gen AI adoption. There’s a wealth of information delivered to the market every day as companies hold earnings or analyst calls and distribute memos, reports, and documents that impact how the market perceives them (changing the stock price in a ruthlessly dynamic way). Going one step further to set up a perpetual feed function to LLMs is the unlock.
Capital markets, by design, are forever-changing. The underlying data that moves markets and funnels to hedge funds, research firms, and financial services organizations is also forever-changing—all in real-time. Traditional hedge fund and finance methods simply aren’t built for that. Gen AI, on the other hand, is built to deliver a forever-updating layer of context on top of data that reflects and supports capital market activities.
