Is permanent capital better suited for AI native services?

AI native services (AINS) don't behave like the startups venture funds were built around, and the difference shows up exactly where fund math lives: in the shape of the return distribution.

Being service businesses, most AINS reach cash flow positive early, often before any institutional round. Once a services company covers its own costs, it rarely dies, so the write off rate that defines venture portfolios (historically around two thirds of deals returning less than the capital invested) mostly disappears. These companies raise to accelerate, funding buy and build strategies like mentioned in my last article or a heavier go to market, not to survive. Take the funding away and most of them would still grow, just slower. The downside case is not a zero but a profitable company on a slower trajectory.

The thing is that the distribution compresses on both tails. Fewer write offs also means fewer 100x outcomes, because services revenue doesn't command software multiples, and a closed end fund built on power law math has no use for what AINS actually produce, which is cash arriving years before any exit window. Permanent capital, on the other hand, is built for exactly that: an evergreen vehicle can collect dividends from the first profitable year, let the compounding run without a forced exit in year eight, and when a company plateaus instead of scaling, sell the stake back to the founders through a buyback at a pre agreed formula. The founders keep a profitable business, and the vehicle recycles the capital into the next one.

Part of the US market has already reached the same conclusion: Thrive Capital is raising a $2B permanent capital vehicle, not a 10 year fund, to run the AI enhancement on service companies model.

A fund should match the shape of the returns it underwrites, and AINS produce cash early, then compound for decades. For this asset class, permanent capital seems to be the logical choice.