The three horizons of AI investment

An AI portfolio with 100% of its investment in operational efficiency is not conservative. It is a portfolio without a future.

The Three Horizons framework (Baghai, Coley and White, The Alchemy of Growth) was not designed for AI. It was designed to manage investment and growth portfolios in any industry. That is exactly why it works so well here: it forces capital to be split across three time frames that compete with each other.

H1 — Optimize the current business

Low risk, high predictability, fast ROI. Forecasting · predictive maintenance · document automation · logistics optimization · internal chatbots.

H2 — Scale new capabilities

Medium risk. This is where sustainable competitive advantage is built. Corporate AI platform · AI Factory · corporate copilots · specialized agents · internal model marketplace.

H3 — Create the future

High uncertainty. Many will never reach the market; a few transform the entire organization. Autonomous agents · cognitive digital twins · scientific AI · proprietary foundation models.

The 70 / 20 / 10 reference

It is not a universal rule (highly innovative sectors allocate far more to H2 and H3), but it works as an immediate diagnosis.

Two patterns show up again and again:

• Everything concentrated in H1: the organization accumulates successful pilots that never scale, because nobody funded the platform that would sustain them.
• H3 overfunded without cash flow to sustain it: that is not vision, it is a bet without backing.

The most frequent mistake, however, is subtler: demanding short-term ROI from an H3 initiative. That is not financial rigor, it is a measurement error, and it is the most efficient way to kill innovation inside a company.

Each horizon is funded differently. And measured differently.

How is your organization's AI investment distributed today?