Neuberger survey finds 78% of GPs raising AI spending on the investment decision itself

Neuberger's two figures count firms, and the gap between them is smaller than it looks. About 90% of the general partners said they were raising AI spending in due diligence and about 78% said the same of the investment decision process, according to Neuberger principal Yi Pan ^[1][2]. The survey as reported does not include budget sizes. So the 12-point gap ^[4] measures how many firms, not how much money: on a base of 140 respondents, roughly 126 firms against roughly 109 ^[5].

The overlap tells you more than the gap. Two groups of 90% and 78% cannot both fit inside 100% unless at least 68% of respondents belong to both ^[6]. The typical GP in the sample is adding AI money at both stages, and diligence is only the more common of the two.

Most of what the Singapore panel described was diligence work. Basil Lui, founding partner of August Global Partners, said his firm's system screens about 2,000 companies a night and narrows the pool for further work ^[7]. Charles Allen of TPG Asia described putting thousands of employee feedback notes through AI while assessing a merger of two businesses ^[9]. "AI not only speeds us up, but also allows us to provide new perspectives on deals," Allen said ^[10].

Staffing plans differ. Lui said AI could do work that once needed a much larger analyst team, with decisions left to senior professionals ^[8]. TPG expects its traditional staffing structure to stay in place ^[12]. "We're now seeing expectations around what each level does and how it gets AI augmented," Allen said ^[13]. At a firm like Lui's, the saving shows up as analysts who are never hired.

I see three ways to read the 78%. It could be spending on committee aids of the kind Pan described, tools that add perspectives while people still vote ^[3]. It could be AI inside the underwriting model, where TPG already counts it as one lever for a company's growth ^[14]. Or, the more interesting version, judgment is moving through the diligence budget, because a screen that cuts 2,000 names a night down to a short list decides which companies a partner ever looks at ^[7].

I think the third reading is where judgment is actually being handed over, and the survey would book that spending under diligence. The other side of the argument is on the record. TPG keeps its tool built on past deals outside the decision-making process ^[11], and Pan said accountability stays with human deal teams ^[19]. If a repeat survey showed the decision-stage share catching the diligence share, that would mean firms are putting AI into the vote as fast as into the research, and my view would be wrong.

The earnings risk sits with the second reading. Xuan Ye of Collyer Capital said he had seen companies with negative returns on AI investments overstate what it would do for customers and revenue ^[15]. Data on some private companies in Southeast Asia is thin (the region where this panel sat) ^[18], and Lui said "Where PE/VC shines is the private proprietary data that the AI can never scrape" ^[17]. Ye was blunter. "AI is not going to bring in good deals," he said ^[16].