In the span of 48 hours last month, Blackstone committed $1.5 billion to a partnership with Anthropic. OpenAI closed a deal with TPG, Brookfield, and others that put another $4 billion to work. Google and KKR were reported in conversations of similar scale. The headlines framed it as a turning point, and they weren't wrong.
But turning points create obligations, not just opportunities. And the governance infrastructure required to meet those obligations was conspicuously absent from every announcement. None of the deals addressed who owns AI risk across the portfolio companies where these tools will be deployed, how AI-informed decisions will be documented and defended, or what happens when something goes wrong.
That gap matters more in this cycle than it would have in previous ones. PE has always made bets under uncertainty, but AI carries a particular kind of unpredictability. The technology is evolving faster than hold periods, the regulatory environment is being written in real time, and the workforce and operational consequences are arriving before the playbook exists.
The reputational pattern is already familiar: value captured in the near term, costs distributed across a longer timeline and onto stakeholders after exit. However, the firms that build durable value from this cycle will be the ones that built the capacity to adapt, defend their decisions, and course-correct when the landscape looks nothing like it does today. That capacity starts with governance infrastructure, and most firms are falling behind.
The questions are already on your desk
LPs are interrogating AI decision-making within standard DDQs now, and the SEC has codified AI governance as a primary 2026 exam priority. For firms with European exposure, the EU AI Act creates immediate, material liability under its high-risk provisions.
Advocacy groups like the Private Equity Stakeholder Project are also indexing PE-backed employers to monitor AI rollouts, placing specific portfolio companies on active watchlists.
None of this is arriving on a future timeline. The questions are already on GPs' desks, and the firms that can't answer them are already behind.
The structural challenge specific to PE
PE firms are typically one step removed from the decisions being made inside portfolio companies. When a portfolio company deploys an AI tool for hiring, scheduling, or customer interactions, the GP typically isn't in the room. Risks are accumulating that the board and investors don’t see.
The incidents that generate LP questions are already happening. AI tools have made hiring and credit decisions on biased or inaccurate information. Customer-facing models trained on proprietary data have exposed that data in breaches. Automated systems have produced confident wrong answers that nobody caught because no review process existed. These aren't edge cases. They're the predictable failures of deploying AI without documented oversight, and many don't have a framework for responding when they do.
The firms positioned to answer these questions are building the infrastructure now. It doesn't require a team of AI ethicists. It requires documented answers to a focused set of operational questions: which AI tools are in use across the portfolio, who authorized them, what decisions they are informing, and who is accountable when outputs cause harm.
Most firms can't answer these questions today.
One question every GP should be able to answer
How are you managing AI risk across your portfolio?
For many firms, the honest answer is still being worked out. That is a reasonable starting point. The technology is moving fast, the regulatory environment is unsettled, and no firm has this fully figured out. But "still being worked out" is only a reasonable answer if you're actually working on it. The firms that navigate this cycle well are the ones building visibility and governance infrastructure now, before an incident forces the conversation.
The AI Risk Readiness Check is built for exactly this. It’s a free resource GPs can use to assess governance infrastructure at the firm level or work through it with portfolio companies. It takes about five minutes and produces results specific enough to act on.
