88% of PE Firms Are Now Spending $1M+ on AI – But Due Diligence Still Comes Down to One Thing AI Can’t Do

Private equity has gone all in on artificial intelligence. According to Deloitte’s most recent M&A generative AI research, 88% of PE firms have already invested more than $1 million into the technology, and two-thirds plan to allocate over a quarter of their budget to AI in 2026. Separate industry data puts adoption even higher: 86% of dealmakers now use generative AI in some part of their workflow, and 84% of firms have gone as far as appointing a Chief AI Officer.

The results, on paper, look impressive. But dig into what the same research says AI still can’t reliably do, and a familiar pattern shows up: firms are automating the reading, not the judgment.

What the Numbers Actually Show

The efficiency gains are real and well documented. McKinsey’s research on generative AI in M&A found that CIM extraction, a task that used to take a deal team 10 to 40 hours, now takes under an hour with AI assistance. Quality-of-earnings review and due diligence overall move about 46% faster. Drafting the first version of an investment committee memo, once a 15-hour task, has been cut to roughly two hours.

Some funds have gone further, compressing entire diligence cycles from the traditional four-to-eight-week timeline down to about six days by moving to AI-native platforms that ingest data rooms, contracts, and financial models all at once.

Where the Data Says AI Still Falls Short

Here’s the part that gets less attention. Even among the firms leading this shift, only about 20% of portfolio companies have actually operationalized generative AI into results that show up on a balance sheet, according to Bain’s Global Private Equity Report. And AI hallucination, meaning the model confidently generating information that isn’t accurate, remains a serious enough concern that multiple 2026 industry reports list it as a top risk in a process where getting a fact wrong can mean overpaying for a company by millions.

This is exactly why expert network calls haven’t gone away, even at firms that have embraced AI everywhere else. Deal teams still rely on live conversations with people who actually worked inside the target company’s industry to validate the assumptions baked into a financial model. What’s changed is only the workflow around those calls. Firms are now using AI to review call transcripts afterward, cross-referencing what one expert said against another, and flagging contradictions before they reach the investment committee. The conversation itself, the part where a real person answers a question AI has no way to independently verify, is still entirely human.

Third Bridge’s 2026 guide to AI-assisted due diligence put it plainly: AI enhances diligence, but it does not replace the human responsibility for capital allocation. That’s not a hedge. Multiple sources describe the same shift happening across the industry: AI accelerates the reading and organizing, while the underlying responsibility for validating an investment thesis, and the accountability if it turns out to be wrong, still sits with people.

A New Regulatory Deadline Raises the Stakes

There’s also a compliance dimension that’s making the human side of diligence harder to skip. The EU AI Act’s enforcement deadline hit in August 2026, and any AI system used in decisions like credit, employment screening, or financial risk assessment inside EU-based portfolio companies now has to meet strict compliance standards. Penalties for violations can reach €35 million or 7% of a company’s worldwide turnover, whichever is higher. For PE firms running AI across a portfolio of EU companies, that’s a direct incentive to keep a documented, human-verified layer in any decision AI touches.

What This Means for Deal Teams Going Forward

The direction of travel isn’t AI replacing human diligence. It’s AI absorbing the document-heavy first pass so that expert calls, interviews, and management conversations, the parts of diligence that actually require judgment, get more focused time rather than less. One 2026 industry guide described this shift directly: by surfacing themes and contradictions in advance, AI makes expert conversations more focused and higher yield, resulting in fewer wasted minutes and more precise validation of the investment thesis.

For deal teams, the practical takeaway is straightforward. Let AI handle the CIM extraction, the first-pass document review, and the memo drafting. But budget real time, and real money, for the conversations that AI still can’t have on its own: the calls with people who actually know whether the numbers in the data room match what’s really happening on the ground.