Beyond MIT's "GenAI Divide": The 6 Pillars to Triple Generative AI ROI

A study by MIT, The GenAI Divide – State of AI in Business 2025 (Aditya Challapally, Chris Pease, Ramesh Raskar, Pradyumna Chari, July 2025), made waves by claiming that 95% of generative AI projects have generated no value over the past two years.

Enough to give more than one executive committee pause — and to call into question the relevance of AI investments.

Rather than stopping at this headline figure, eleven and Revenu-IA wanted to go further: to understand what the study actually says, its methodological limitations, and above all to identify what really distinguishes, in the field, the projects that fail from those that genuinely scale.

A clear objective

👉 Move beyond the shock number, and formalize the 6 pillars that make it possible to achieve 3x ROI in generative AI within 12 to 18 months.

Key figures

• MIT study based on 300 projects analyzed and 150 executives interviewed
• 60–70% "at scale" success rate among eleven & Revenu-IA clients
• ROI often reaching 3x within 12–18 months relative to the initial investment

What you'll find in the report

A structured analysis across 6 pillars, illustrated with concrete client case studies:

1. What MIT's study really says (and its methodological limitations)
2. A dual "Business & Tech" approach from day one
3. Deliberate and continuous selectivity
4. "Off-the-shelf solution"… or not: when custom-built is cheaper and more relevant
5. Winning over business teams by starting from their real challenges
6. End-to-end team continuity, "from design to code"
7. Bridging the skills gap to align executives, business teams, and technical teams

‍