Proserv & PE Insights · Friday, October 2, 2026

Hi, this is Koko from Koko Knows.

Let's start with the sharpest thing I've seen land, and it's coming straight from inside the room. A PwC partner has gone on record saying clients now expect projects delivered twice as fast because of AI. That's not a vendor promising speed, that's a partner admitting the client side has already reset its expectations and the firm is scrambling to keep up. This is the clearest evidence yet that fee compression isn't a future risk you're modeling, it's happening in live engagements right now, and firms haven't figured out how to reprice fast enough to protect margin.

Sitting right alongside that is a pair of new Bain reports that should worry anyone with AI exposure in a portfolio or an advisory book. Bain is pointing to a gap of four to almost five trillion dollars between what the industry is going to spend on AI infrastructure, roughly one and a half trillion dollars a year by 2031, and the software revenue that's supposed to justify it. In plain terms, the ROI story that's been underwriting valuations, consulting mandates, and vendor pricing may simply not close. If you're sitting on a portfolio company whose multiple assumes AI-driven upside, this is the number to stress-test against before your next board meeting, not after.

Then there's BCG, which confirms something a lot of us have suspected for a while but hadn't seen formalized: the shift to outcome-based pricing is now a confirmed trend across the top consulting peers, not an experiment. That matters because it's the industry quietly admitting that hourly billing on junior staff no longer works once AI tools are doing the junior work. Deloitte backs this up with hard numbers, estimating thirty to thirty-five percent productivity gains from AI coding tools across the software development lifecycle. That's not a marginal efficiency story, that's enough to make clients walk into a renewal conversation and demand a lower build cost or a faster timeline, full stop.

And then KPMG adds the fourth leg here with their new Global Tech Report: sixty percent of organizations say their AI investment is outpacing their governance capacity. That's a massive, validated opening for advisory fees around AI risk and readiness, and frankly it's one of the few growth lines in this whole picture that isn't under pricing pressure. If anything, it's the inverse, clients will pay a premium to not blow something up.

So what does all this mean if you're a partner or an MD, or sitting on the investment committee side? First, speed expectations are now outrunning your ability to reprice fairly. That PwC admission is a warning shot. If you haven't already moved your major engagements onto outcome-based contracts, you're exposed to clients who've already decided, unilaterally, that projects should take half the time, and they're not going to wait for your billing model to catch up. Lock in outcome-based pricing now, while you still have leverage, because BCG's data says the rest of the market is already moving there.

Second, treat the Bain gap as a real constraint on how aggressively you underwrite AI-driven value creation stories. If the industry-wide revenue math doesn't support the infrastructure spend, any single portfolio thesis that leans heavily on AI margin expansion needs independent verification, not vendor assurance. That's especially true given what we're seeing elsewhere this week, with Anthropic's own leaked financials showing billions in losses against a fraction of that in revenue, and hardware players like Samsung and Apple already passing memory shortage costs down the chain. Unit economics in this space are provisional until proven, not something to assume into a deal model.

Third, and this is the opportunity side, the governance gap KPMG is flagging, and frankly the broader pattern of agentic tools shipping faster than anyone can govern them, is real fee-generating work. Boards should be asking for a governance timeline before any large-scale agent deployment, and that's a conversation professional services firms are uniquely positioned to own, assuming you build the practice now rather than after the first high-profile failure.

Last thing worth flagging for delivery risk: Gartner's forecasting that seventy percent of vendor-led forward-deployed engineering models get abandoned by twenty twenty-eight, and separate research shows seven in ten tech leaders already scrapped an AI project this year due to talent gaps, at a quarter million dollars or more in wasted spend each time. If your services or GTM motion still depends on that heavy, bespoke delivery model, this is the moment to start migrating toward something more durable and self-service, before it shows up as a liability in your own pipeline.

That's the state of play. Margin, governance, and contract structure are all moving at once, and the firms that reprice and build the advisory lines first are the ones that come out ahead. I'm Koko, and this has been Koko Knows.