New Roles, Diverging M&A Trends, and Lagging Governance.

There's been a growing narrative that M&A is "back". While this is technically true, the details matter, and we're going to take a closer look here.

The headline statistic is that global M&A deal values increased ~40% in 2025 (PwC states a 36% jump to roughly $5 trillion; McKinsey a 43% jump to $4.7 trillion).

But deal volume — the number of transactions getting done — was essentially flat, according to both PwC and McKinsey. So more accurately, we could say that M&A megadeals are back, and there's good reason for this.

What's causing the rebound in larger transactions specifically? There are three identifiable trends that can help explain this movement.

First, AI infrastructure is pulling capital upward. As we discussed last week, AI transactions are on the rise — and a large number of these transactions are megadeals by necessity. Equity tickets to compete in data center capacity or model infrastructure are almost exclusively measured in the billions — CoreWeave bid $9B for Core Scientific, and BlackRock paid $40B for Aligned Data Centers. When AI infrastructure dominates M&A activity, deal values are naturally pulled upward. To further enforce this point, PwC found that roughly one-third of the 100 largest M&A deals in 2025 referenced AI in the strategic rationale, and the largest technology deals almost invariably cited it specifically.

Second, private equity sponsors are changing the way they deploy capital. PE deal value jumped to $1.2 trillion in 2025, but total deal count declined year over year. We can infer that firms are pursuing the efficiency of fewer, larger deals, rather than applying capital and resources across numerous, smaller transactions. Interestingly, at the same time we are seeing average hold times increase to 6.2 years, versus just four years in 2009. This would also suggest that a lot of firms are going to be increasingly focused on exit activity in the coming months and years.

Third, tariff uncertainty is a major driver of deal hesitation, particularly in the mid-market. In the first half of 2025, mid-market activity was noticeably suppressed, with many buyers and sellers adopting a "wait-and-see" mentality — a frame of mind that anyone working in corporate development can identify all too clearly when dealing with uncertain founders. PwC noted that tariff policy shifts made forecasting especially difficult for mid-sized firms, with CEO confidence scores cratering accordingly.

By contrast, many of the megadeals that did go through were largely part of complex strategic moves where the big-picture theses like AI infrastructure and consolidation overrode macro noise. In other words, transactions that were either too big to stop or too costly to delay.

Fortunately for us in corp dev, signs are generally pointing towards yes, gradually. Remember, these businesses aren't disappearing — they are waiting. The narrative we've all been following over the last five to ten years, of succession planning, aging wealth class, and the need for mid-market buyers has not changed, but the business owners are adapting to the speed bumps in the overall economy.

A 37-year Capstone Partners veteran recently said 2026 "looks legitimately promising", pointing to private credit dominance, PE deployment pressure, and improving buyer-seller alignment as indications of mid-market resilience. Another thing to consider, and particularly if you've worked for any of the number of Vertical Market Software consolidators, is the increasing pressure — both internally and externally — to deploy capital. These transactions tend to occur in the mid-market.

The market is opening, but it's not opening to everyone equally.

SS&C Intralinks released their report in collaboration with Reuters last week, titled "AI in Dealmaking: A Benchmark Study". The report was a survey of 400+ global M&A dealmakers pertaining to their use of AI in the course of deal work.

Adoption in one form or another is all but ubiquitous. 90% of dealmakers have integrated AI into some part of their process, and other surveys (Axial, Bain, Accenture) are showing similar data. Pre-deal activity, such as screening and diligence, is showing a marked increase since 2024. In other words, firms are unanimously agreeing that AI is additive to their M&A processes.

But widespread adoption does not always come with safety checks and proper governance. 80% of those same dealmakers reported experiencing AI-related security incidents or near misses in the last year. The two largest sources were access-control lapses, accounting for 48% of reported incidents, and hallucinated outputs leading to inaccurate diligence, accounting for 40% of reported incidents. This in particular is striking — almost half of the firms surveyed have had an AI tool produce fabricated information during a formal diligence process.

If you want a real world example, look no further than the Deloitte AI scandal, which broke headlines in Australia and around the world back in October of 2025. Australia's Department of Employment and Workplace Relations had engaged Deloitte to produce an independent assurance review of the IT systems used to automate welfare compliance penalties. The report featured citations of academic studies and policy papers to strengthen the findings — but there was a structural problem rooted in the research foundation. An independent academic later reviewed the report and found that several cited papers didn't exist, and that references made in the report couldn't be traced back to any journals or databases. Deloitte had used AI to help create the report, and the AI had fabricated source material out of thin air. Deloitte was forced to repay part of the ~$440,000 AUD contract, and had to withdraw and rewrite the report.

Cases like this show the dangers of over-reliance on AI without proper governance. Firms using AI in deal diligence should be rigorously testing any assumptions and reviewing critical inputs to reduce the risks of AI hallucination. On one hand, firms need to remain cautious about the risks of AI, but on the other, they must continue adoption in order to remain competitive with their contemporaries.

The real question is whether AI governance can evolve fast enough to keep up with adoption, or whether AI will continue advancing at a pace that makes effective regulation nearly impossible.