If the revenue engine is messy, AI scales the confusion
Hello there!
It seems fashionable at the moment to be an AI futurist on a podcast. Almost any version of the future, fantastic success, doom and gloom, somewhere in the middle, seems to attract a very eager audience. Humans are typically not great at dealing with either uncertainty or exponential change and it seems like both are in the works. More on just a small slice of what I've been listening to below.
I remain convinced that access to proprietary first-party data is going to become even more critical in the success of a company. Assuming we'll all still be around to run companies.. 😄
Busy week this week as conference and travel season is about to kick into full swing - starting with ACG DealMax in Vegas next week, let me know if you're going to be there and want to connect!
Enjoy the rest of this week's memo!
Three Things I Learned This Week
The new CRO is a revenue underwriter, not a pipeline manager
This week’s PE Data Guy episode is with Derek Sather, CRO at Education Perfect, a KKR-backed edtech company based in Sydney. Derek spent six years at Winning by Design helping 600+ software companies stop guessing about revenue and start engineering it. He is MIT-trained and thinks about revenue the way an actuary thinks about risk.
Three of his lines I am going to keep stealing.
- “One metric, one owner, one logic chain. Once that number is trusted, you move on to the next one. That’s how you build credibility. Not with a huge giant transformation story. One defensible number at a time.”
- “Your revenue number is critical. If you can’t trace it to source systems, it’s not an asset. It’s just a claim.”
- “A messy revenue engine can still hit plan for a while. It just can’t survive scrutiny.”
The framing he landed on by the end of the conversation is the one worth sitting with. The old CRO was a pipeline manager. The new CRO is a revenue underwriter. The job is no longer just to produce growth; it is to distinguish good growth from bad. Can this revenue close at top pricing and valuation? Are we pulling demand forward? Are we masking churn? Are we creating problems for the next owner? That is the lens portfolio CEOs should be applying three years out from an exit, not three months.
The other thing worth stealing is the compounding math. A 10% improvement on each of seven conversion points doubles revenue. The British cycling team winning gold medals from 1% gains on bicycle weight. F1 pit stops compressed from 20 seconds to under 2. Revenue engineering is the same discipline. The operators who understand this are building investor-grade revenue systems quietly while their peers chase the next big transformation story.
Watch the full conversation here.
AI is going to the edge, and the moon might follow
I spent a chunk of time this week listening to David Friedberg on Modern Wisdom with Chris Williamson.
AI is not going to concentrate in mega-corporations. It is diffusing. Cost per token has dropped a thousandfold. Open-source models run on a MacBook. Foundation capabilities are heading to desktops, phones, and embedded devices. The implication for PE is not theoretical. If compute for useful AI is effectively free in three years, the competitive moat on any portfolio company that was betting on “we have AI and the small guys do not” evaporates. The moat has to come from data, relationships, process rigor, and domain knowledge, not from access to models.
The full conversation is here if you want to go deep, including a discussion of the moon as humanity's future industrial hub!
PE is hedging against AI disruption by flooding into the trades
Oak Hill Capital bought Guild Garage Group earlier this month for $800M at 16x EBITDA. 16x is residential services priced like an IT or data services company. 26 PE-backed deals in the garage door sector in 2025. 29 in 2024. Both years dwarfed what came before.
Why is this size of multiple is showing up in garage doors at all? One PE partner quoted in PitchBook put it plainly. “Some of the historical perspective on blue collar services and just construction, as being kind of a dirty word in private equity, that’s gone.” The read is that PE firms historically focused elsewhere are moving into the trades as a hedge against AI-disruption exposure in their other portfolio companies. Fragmented, cashflow-generative, mission-critical services. Hard to automate away. Easy to consolidate.
Two News Stories From This Week in Mid-Market PE and Data
156 PE deals in a single month. The window is open right now.
Sources: Infocapital "PE Acquisitions Surge" (April 18, 2026) | McKinsey Global Private Markets Report 2026
What happened. April 2026 is on pace to be the most active PE acquisition month in over a decade. 156 closed deals in the first 20 days, $123B in disclosed transaction value. Forty-six PE-backed deals were announced on April 10 and 11 alone, including Hologic at $18.3B, Sealed Air at $10.3B, and CVC's €10.9B move for Recordati.
Annualize that pace and you get 1,872 transactions for the year, higher than the 2021 peak when dry powder hit record highs. The timing is not a coincidence. KKR closed a $23B North America fund earlier in April. Blackstone, Apollo, and Carlyle all closed mega-funds in March. Firms with fresh capital are deploying on cue, and PitchBook data shows the mega-deal concentration is masking a deeper shift: the mid-market is flowing again too.
Why you should care. For 18 months, the market told mid-market PE operators that patience was the right call. Hold, improve, wait for the window. The window appears to have opened in April. Whether that turns into a quarter or a year is unknowable, but the pattern after a multi-year drought tends to favor the first movers. The problem is that the portfolio companies that spent the drought on cost discipline, not on data, are the ones who will discover in diligence that they are not ready for the exit they are now competing for.
Buyers doing diligence in this window will see 10 other targets at the same time. The ones with reconcilable revenue, defensible customer counts, and a credible data room trade. The ones that cannot answer the third follow-up question lose a turn on the multiple and wait another cycle. The market is not waiting for you. The diligence team sitting across the table is doing simultaneous diligence on three other companies right now. The data readiness gap has never mattered more.
OpenAI’s GPT-Rosalind moves frontier AI into biology
Sources: PitchBook “Frontier labs come for biology” (April 20 edition)
What happened. OpenAI launched GPT-Rosalind, its first model purpose-built for life sciences. VCs had largely bet that frontier labs would leave biology alone on the assumption that drug discovery required proprietary data and specialized architecture that generalist foundation models would not bother to build. That bet is now wrong. “Life sciences have moved from being a side interest to a top priority for leading AI labs,” per Sofinnova Partners. PitchBook’s AI analyst called it bluntly: “Rosalind is OpenAI entering a race Google is already running.” DeepMind’s head start on domain-specific biological reasoning is still the one to watch.
Why you should care. If you have biotech, pharma services, diagnostics, or life-sciences-adjacent companies in a PE portfolio, the competitive landscape just changed. Specialized AI-for-biology startups now have to compete with frontier labs for the same pool of talent, the same compute, and eventually the same enterprise customers. The portfolio companies that win the next cycle are the ones that already have the proprietary data assets to fine-tune on top of whatever model wins. The ones that were planning to build their own models from scratch are now in a tougher market. It is the same pattern as every other vertical AI wave. The data asset is the moat. The model is a commodity you rent.
Free Tool of the Week - Data Room Survivor
You are the CTO. The buyer’s tech team just landed. Ten scenarios, five minutes, and every decision you make affects confidence, timeline, and valuation. Data Room Survivor is a free interactive simulation of the 10 days of diligence that decide whether your data holds up or whether a multiple comes off the price. Several readers have told me it was the most honest mirror they have held up to their own data room in months.
Sign-off
If any of this lands, and there is something you think we can help with, just reply. We read everything that comes through.
As always, forward this on to your favorite PE-backed friend.
Cheers,
Graeme