40% of PE firms walked away from a tech deal over AI disruption.

Over the past year, 40% of private-equity firms in EY's latest pulse survey say they stepped away from a potential technology investment because of AI-led disruption.

That figure landed on October 29 in EY's Private Equity Pulse for the third quarter of 2026, written by Pete Witte, EY's global private-equity lead analyst. The buyers did not swear off software. They walked when they could not underwrite what the company would still own once a model could do part of the job.

When AI risk showed up in a live assessment, 46% of the firms said they expanded diligence or revised the investment thesis. About a third renegotiated the price or cut the proposed valuation. About a fifth changed the deal structure. Walking away sat next to those softer moves, not above them as a panic number.

For a founder waiting on a sale, or a household whose wealth sits in a services shop, the useful question is concrete. What does the company do that a model cannot cheaply copy?

Andrew Collins, a managing director who co-heads the technology team at growth-equity firm Summit Partners, put the test in four questions for Private Equity International. Does the software serve as a system of record? Does it hold proprietary data? Could a customer fail an audit if the product failed? Are the error costs high? Those are the traits he calls defensible.

Code alone answers less of that test than it used to. AI-assisted development makes it cheaper to rebuild a feature. What still holds a price is the workflow customers already trust, the data competitors cannot buy, and the places where a wrong output is expensive.

India's IT-services industry long sold engineers times billing rate times utilization. Manisha Girotra, chief executive of Moelis India, has said that equation is under pressure as routine work automates and clients want implementation value, not headcount. Reuters reported in August that large Indian IT firms are shifting more contracts toward outcomes and watching some work move back in-house.

Aalok Shah, who covers technology deals for Rothschild & Co in India, has described the same pause on the buy side. Private-equity sponsors are finding it harder to underwrite generalist technology-services platforms, while specialized vertical shops still draw interest. A bench of interchangeable hours looks different from a regulated workflow a hospital or bank cannot casually swap.

EY's pulse is a small, directional sample, not a census of every deal. AlphaSights fields the online survey each quarter among roughly 50 private-equity professionals at the VP level and above, at firms with at least $1 billion under management across the Americas, Europe, the Middle East, Africa, and Asia-Pacific. The 40% is a sentiment check. It is not the odds on one term sheet.

Inside portfolios, the same firms are spending more on AI. 80% of general partners said AI-related spending across their companies rose over the past year. For 39%, the increase was between 25% and 49%. Half pointed to software models, APIs, and vendor licenses among their top investments. Cloud, compute, and workflow redesign followed.

The returns are still early. 40% said benefits have shown up but have not been quantified in dollars. 27% said they have too few live projects to judge. Only 19% said they are already seeing positive net returns. A buyer can fund AI tools inside a company and still refuse to buy a company whose product AI can replace.

A separate EY exit-readiness study earlier in 2026 found the share of general partners who call AI a significant challenge when preparing a company for sale rose from 7% to 18%. That report said buyers now separate a stack of pilots from a real plan: where AI sits in the operating model, how the benefit is measured, and what happens under the next owner.

If you are the seller, or you sit on a PE-backed bonus, diligence comes back to Collins's questions. System of record. Proprietary data. Audit trail. Cost of being wrong. A pitch built on more people writing the same code a model is learning to write is the one that gets left on the table.

If your compensation sits in RSUs at a software employer navigating the same questions, RSU Vest Day Playbook for Diaspora Households keeps vest math separate from the job story. Career exposure by industry is in Young employment in AI-exposed industries fell 12%. Quants and medicine still pay.. Founders thinking about a sale years out can keep Succession and Exit Planning for Immigrant-Owned Family Shops nearby even when the "shop" is a SaaS book, not a storefront.

The firms that walked were not arguing about whether AI exists. They were arguing about whether the company still had something left to sell.