Investment firms have spent big on GenAI over the past year, yet the returns aren’t landing evenly. Some companies are seeing real ROI and making better-informed decisions, while others are seeing little for the same spend.
According to Clearwater’s CEO, Sandeep Sahai, budget and tools deployed don’t explain the gap. Data accuracy does.
Clearwater set out to test that this year, polling senior leaders at insurers, hedge funds, private credit firms and asset managers. The findings detailed in GenAI and the Data Divide showed it all came down to whether firms can actually trust their data.
Speaking with Fintech TV, CEO Sandeep Sahai connected that finding to a broader shift now underway across the industry.
Foundation models are normalizing, he said. A year ago, the conversation was all OpenAI. Now it’s Anthropic. Compute is normalizing too. Nobody thinks twice about what chip is running their iPhone anymore, the same way nobody asked in the 386 era. When the model and the hardware stop being the differentiator, what’s left is whether you can believe your data.
That’s exactly what the survey showed. 79% of firms rate their data as complete, yet only 56% rate it as accurate. That’s a 23-point gap between having data and trusting it. And 99% of executives now name unreliable data as the single biggest barrier to trusting what their AI tools produce.
The gap shows up in performance, not just perception. Among firms with strong data accuracy, 44% say their risk management has become much more proactive. Among firms with weaker data, that number drops to 17%. Same tools, same investment, very different outcomes.
Sahai pointed to what those tools can do once the data underneath them is solid. Take a cotton futures trader. Instead of making a few calls to gauge next year’s crop, 1,000 agents can swarm every zip code that grows cotton, tracking soil temperature and rainfall in real time. Or take a personal savings account. Most people check the rate once, at signup, and never again. A hundred agents can watch every bank persistently and flag the best return the moment it changes.
The same principle applies at industry scale. Insurers have grown their private credit holdings, with the Credit Where It’s Due report showing this rose from 5% to 9%. Wealth managers have gone from holding none to holding 5%. None of that works without credible data behind opaque instruments such as private credit, CLOs, and mortgage-backed securities.
That’s the foundation Sahai keeps coming back to. Spend and tenure don’t predict which firms get more out of GenAI. Accurate data does.
With an accurate dataset, firms can see investment opportunities worldwide, in India, Tokyo, or Europe, clearly enough to act without taking unnecessary risks. That’s the bet Clearwater is making, and the one Sahai thinks every firm serious about GenAI will eventually have to make too.
Explore Clearwater’s GenAI capability.
Read the full findings in GenAI and the Data Divide.
See the full interview on Fintech TV.