Your Competitors Just Locked In AI at Fund Level. Now What?
This is the second follow-up to my post on the PE/AI joint ventures forming between major funds and AI labs. The first covered what it means for CTOs inside portfolio companies. This one is for the operating partner or technology partner at a PE firm that has not signed one of these deals, and the question is simple: what do you do when your competitors have locked in preferential AI access across their portfolios and you have not?
TPG and Bain Capital have a joint venture with OpenAI; Blackstone and Permira are doing the same with Anthropic. These are multi-billion-dollar commitments with guaranteed returns and mandated deployment across hundreds of portfolio companies. Their portfolio companies get AI infrastructure at scale, with fund-level support, preferred pricing, and dedicated integration resources. Your portfolio companies are still evaluating vendors individually, running pilots that stall at month three, treating AI as an operational line item rather than a structural capability. The gap compounds because AI gets better with usage; the portfolios that deploy first generate production data, refine use cases, and build institutional knowledge that you cannot replicate by signing a contract six months later.
I have seen how most PE firms handle cross-portfolio technology initiatives, and I have been inside enough of them to know the pattern: an operating partner runs a knowledge-sharing session across portfolio CTOs, someone suggests a preferred vendor list, a few companies run pilots, most do not, and a year later the initiative quietly disappears from the board deck. That approach will not close the gap here. The funds with joint ventures are deploying AI as infrastructure across their portfolios with capital commitment and board-level accountability behind it, and a knowledge-sharing call is not a competitive response to that.
You are not locked out, but you need to move with intent rather than the usual committee-driven pace, and there are three approaches worth considering. The first is to build your own fund-level AI capability without a joint venture. You do not need a deal with OpenAI to build something credible; what you need is someone who can assess each portfolio company's readiness, identify the highest-value use cases, and run deployments that actually reach production. I have done this work at the portfolio company level — assessed data architecture, built production AI systems, deployed them into regulated environments — and the same approach works at fund level with coordination across companies and a clear view of where AI creates real value versus where it is theatre. The advantage is that you choose the right AI tools for each company's specific needs rather than mandating one platform across businesses with very different tech stacks and data maturity levels.
The second option is to use AI as a due diligence weapon. If you are not ready for portfolio-wide deployment, start with where AI creates immediate, measurable value: the deal process itself. I wrote about this in my post on AI value sitting upstream of engineering; the same capabilities that accelerate code migration can compress technical due diligence dramatically, comparing feature coverage across codebases, quantifying technical debt, mapping integration complexity, and estimating modernisation effort. What took three months of manual analysis I have seen compressed into two weeks. This does not require portfolio-wide readiness, it delivers value on every deal from the moment you deploy it, and it builds internal muscle around AI that you can expand later.
The third option is to target the second wave. The first wave of these joint ventures will produce lessons; some deployments will succeed and others will stall because the portfolio companies were not ready — data was a mess, teams lacked the skills, governance was non-existent. The AI labs will learn what works, and the deal structures will evolve. If you cannot move now, prepare to move fast when the second wave opens by auditing data maturity across your companies, identifying which businesses have the infrastructure to absorb AI and which need foundational work first, and building a shortlist of use cases by company. When the next round of deals surfaces, you negotiate from a position of readiness rather than scrambling to catch up.
The worst response is the passive one — telling yourself it is too early, or that the joint ventures might not work, or that your portfolio companies can handle AI adoption on their own timelines. They cannot, not at the pace required to compete with mandated, fund-level deployment. The second worst response is panic buying — signing up for an AI platform at fund level without assessing whether your portfolio companies can actually absorb it. I have written about what breaks when AI platforms land in environments that are not ready, and deploying AI into unready companies wastes money and damages credibility with management teams.
This is fundamentally a portfolio value creation question. The funds with joint ventures are betting that AI deployed at scale will generate enough EBITDA uplift to justify the investment, and they are building a structural advantage that compounds over time. Whether you can build an equivalent advantage without the joint venture structure depends on taking a deliberate, execution-focused approach rather than the usual committee-and-preferred-vendor-list playbook. The starting point is honest assessment: which portfolio companies are ready, where does AI create real value rather than demo value, and who has the data maturity to support it. Answer those questions and you have a plan; ignore them and the gap keeps widening.