Harvard Business Review (January–February 2026 issue) · Article
Match Your AI Strategy to Your Organization's Reality
GM's AI-designed seat bracket was 40% lighter and 20% stronger, and it never made it into production.
- Published
- Reading time
- About 18 min
- Authors
- Cyril Bouquet, Christopher J. Wright, Julian Nolan
What is it about?
The article opens with two 2018 bets. GM used generative design in Autodesk Fusion 360 to create a seat bracket that consolidated eight parts into one, 40% lighter and 20% stronger, but its manufacturing system, built for stamped steel, could not produce it at scale. Apple's metalens work reportedly reached the Face ID system because Apple controlled the whole stack around it.
The authors argue that most AI failures come from a mismatch between ambition and what the value chain, operating model and tech stack can support. Their 2×2 maps value-chain control against technological breadth, giving four strategies (focused differentiation, vertical integration, collaborative ecosystem, platform leadership), each with a typical way to fail.
Key numbers
- GM bracket: 8 parts became 1, 40% lighter, 20% stronger, never mass-produced.
- S&P Global: 42% of companies abandoned most of their AI initiatives, up from 17% a year earlier.
- A Writer survey: 31% of employees admit to working against their company's AI strategy.
- Named failures by quadrant: Zillow Offers, GE Predix, IBM Watson at MD Anderson, DeepMind and the NHS.
Why read it
- It offers a simple test before you pick use cases: what do you control end to end, and how many technologies must you combine?
- The GM story is the cleanest illustration that a better AI output can be worth zero if the organization cannot absorb it.
- It is good pre-mortem material, since each quadrant comes with a named failure to check against.
Who should read it
- CIOs building a multi-year AI roadmap
- Strategy and innovation leaders
- Manufacturing and supply-chain executives evaluating AI-designed products
About the author: Cyril Bouquet
Professor of Strategy and Innovation, IMD Business School
Bouquet studies why good ideas die inside organizations. His award-winning PhD became the book “Building Global Mindsets”, and he co-wrote “ALIEN Thinking: The Unconventional Path to Breakthrough Ideas” (2021).
He designed the “mega dive”, a large-group transformation format he has run with 700 senior leaders at Maersk during its shift from shipping to integrated logistics and with the top 650 leaders of France's SNCF. Before IMD he taught strategy at York University's Schulich School of Business in Toronto.
Read it carefully: common misquotes
- GM's bracket used algorithmic generative design, not LLM-style generative AI, and GM always presented it as a 3D-printed proof of concept.
- Apple has not confirmed the metalens; the evidence is an analyst teardown of the iPhone 17 Pro Max.
- Two of the three authors are executives at Iprova, an AI invention company, and featured example P&G is an Iprova client. The article discloses this.
How it shapes the CIO's AI Playbook
- Step 2.3: Check that the organization can absorb the result · Choose where to play
Notes by Pavel Veselov, last reviewed . Summaries are my own; quotes and figures belong to the original authors.
All 11 sources in the CIO's AI Library