McKinsey partners time a corporate venture's first $10m at 31 months

The compression amounts to seven months: 38 months to a venture's first $10 million of revenue, now 31, or 18% off the clock ^[1][1]. The capital claim is bigger than that. Breaking even on 40% less money while taking 31 months instead of 38 means spending per month fell too. 0.6 times 38, divided by 31, is 0.74, so the average venture burns about 26% less a month than its predecessor did ^[2][2]. A bit under half of the 40% comes from simply finishing sooner ^[3].

Whether the build option got cheaper or only quicker decides how much a CFO can knock off a purchase price. The McKinsey partners writing in Fortune attribute both halves to AI: lower experimentation costs, quicker build cycles, and businesses designed as AI-native from the outset ^[3][12]. Their article does not say how many ventures the averages cover or over what period they were measured ^[13].

Success defines the sample. Cheaper experiments let a company run more bets, so the winners are drawn from a bigger pool and the average winner looks faster and thriftier even if no single venture got cheaper to build. Capital spent on the bets that were stopped does not appear in a break-even average measured on the ones that worked. The article's own advice runs the same way: three or more simultaneous ventures for up to 30% higher revenue growth than a single bet ^[5]. It also calls for short review cycles and clear funding thresholds so a CEO can call a halt ^[11].

Each case is offered as a ratio against a parent, and a small base grows faster than a large one. Honeywell Connected Enterprise reached around $1.5 billion of annual sales by 2023 and is growing about three times faster than Honeywell overall ^[6]. STC Group's subsidiaries in payments, internet of things, cyber security, data centers and cloud infrastructure are growing 10 to 15 times faster than the core business ^[7].

What the partners call the catch sits inside the company. As the venture succeeds, the governance, teams, processes and controls that run the parent arrive with it, and they say the CEO has to step in at that point ^[9]. BCP's answer was to staff Yape with product development, engineering and design talent instead of traditional bankers, and the wallet has since passed 18 million users ^[8]. The corporate parent supplies customers, capital, data, expertise, distribution and a brand; the difficulty is taking them without the constraints attached ^[14]. Leadership is asked to read commercial evidence, because a venture can hit every project milestone and still fail to become a good business ^[10].

In my view the 31 months is the usable number, since it is measured the same way at both ends and sets a horizon a CFO can discount against. The 40% is the softer one, because it depends on which ventures entered the average. The counter-thesis is that AI raised the number of attempts, leaving the survivors looking thriftier while the cohort spends as much as ever. What would settle it is total capital deployed per success across a full cohort, stopped bets included. If that has not fallen, the 40% describes survivors, and the build leg of buy, build or partner costs what it always did ^[4].