Moatti prices counter-positioning at twice what proprietary data earns across 576 AI rounds

The multiple's denominator is money raised. Double the capital at an unchanged enterprise value and the multiple halves ^[7]. So the 5.3x on counter-positioning and the 2.6x on cornered resources ^[4][6] rank how little a company needed alongside how well it is defended. Per dollar raised, counter-positioning prices at about 2.04 times cornered resources ^[1], and about 1.26 times network economies ^[8].

The subgroups are small. Five percent of 576 is roughly 29 companies ^[2], so the top multiple in the analysis is a median across about 29 firms, while the 44% in cornered resources is about 253 ^[3]. Every company in the set closed a round of $50m or more since the start of 2025 ^[1]. At that floor the sample holds at least $28.8bn of capital ^[4], with about $12.65bn of it sitting in the category the analysis scores worst ^[5].

Two orderings in the piece do not line up. Cornered resources at 44% is called the second-highest prevalence in the dataset, and switching costs at 37% is called the most crowded power ^[6][7]. The article does not disclose a switching-costs multiple, or whether one company can be counted under more than one power ^[14]. Take the four disclosed shares at face value and 10% of the sample sits in the two powers SC Moatti endorses against 81% in the two she calls traps ^[6].

Moatti wrote in Crunchbase News that a pitch still leading with the use of AI is describing infrastructure, not a business ^[8][15]. Her case against the crowded 44% is that investors have watched too many proprietary datasets get eroded by foundation models and synthetic data. On her account, a data advantage which does not compound in ways that get harder to replicate over time is not a moat ^[12]. Her case against switching costs is cost of construction: baked-in instrumentation, institutional memory and rearchitecture risk mean expensive sales cycles first ^[13].

Counter-positioning, in her account, is a bet on someone else's income statement. Blockbuster could have matched the subscription model but doing so would have gutted the late-fee revenue that kept its stores alive, so it did not, until it was too late ^[11]. "The incumbent sees the threat. They choose not to respond. That rational inaction is the moat," Moatti wrote ^[9]. The diagnostic she offers founders is whether a well-resourced incumbent could copy the model, with the right answer being "technically yes, but it would cost them more than it would cost us to build" ^[10].

I would put more weight on the 4.2x than on the 5.3x. Network economies are countable at the time of the cheque: participants on each side, and whether the second side committed. That is the cold-start problem she says a better-funded rival with a better model still cannot buy ^[17][5]. Counter-positioning is a forecast about how a competitor will price its own cannibalisation, and it is confirmed only after that competitor declines to respond ^[9]. The counter-thesis is that enterprise value per dollar raised is the number a venture buyer actually transacts on, so a 2x gap in entry price ^[1] is real money whatever caused it.

What would settle it is the next round. If the roughly 29 counter-positioned companies ^[2] hold 5.3x on a larger denominator while the 253 in cornered resources ^[3] slip under 2.6x, the classification predicted something. If the 44% re-rate upward as their datasets compound, the 2.6x was a 2025 funding-market price and not a verdict on the business.