37% of New Directors Are CEOs as Boards Build for the Last War

Current and former CEOs took 37% of new board appointments in 2026 — the highest share since 2012 — while fewer than 3% of S&P 500 directors disclose any AI expertise. That's the gap at the heart of the latest governance warning: boards are staffing for the last crisis while the next one arrives from a direction nobody's watching.

More than four in five S&P 500 companies now disclose AI as a risk, according to the Conference Board. The director data comes from Spencer Stuart's 2026 Board Index. Put the two together, per the analysis, and boards risk building a Maginot Line: France's 200-plus-mile fortification network, designed for the static warfare of World War I. Germany's blitzkrieg bypassed it entirely in May 1940, breaking through French lines within five days of launching its attack on May 10.

The author's read is blunt: boards can see where the next threat is coming from, yet they are still staffing for the last one.

Why are boards doubling down on experience?

In 2026, retired executives made up 54% of new directors. The logic is understandable — in tough times, you pick people who have seen it all before. A CEO who has lived through a recession has already paid for lessons the rest of the board doesn't have to learn the hard way.

But experience is, by definition, a record of past mistakes. Oscar Wilde's line — "Experience is the name everyone gives to their mistakes" — cuts both ways. It protects against the previous era's errors, not the next era's. The lesson of 1940 isn't that France should have skipped the fortifications; the mistake was relying so heavily on them. Experienced directors still matter for CEO succession and crisis management. Overemphasize them, though, and the board loses sight of how AI is actually entering the business — as something that upends industries, org charts, and the nature of work, not just an IT upgrade.

Is this a generational problem?

No, and reading it as one misses the point. Age is only a proxy. A 62-year-old executive who works with AI daily is closer to the present than a 45-year-old who left an operating role five years ago. The analysis proposes three tests instead.

Recency. Are they working in today's conditions? Consider Matías Muchnick, cofounder and CEO of Chilean food-tech company NotCo. When cocoa prices spiked in 2024, Ferrero turned to NotCo's AI to find substitute ingredients across its more than 30,000 annual recipes. Muchnick saw the technology had become more valuable than the products: he split off the AI business and, this year, sold food operations in Argentina, Uruguay, and Brazil. That fluency is hard to acquire secondhand. First step: dedicate one seat in the next refreshment cycle to a director with current operating experience in the technologies reshaping the industry — or create a term-limited observer role.

Proximity. AI arrives through employees and customers long before policy catches up. In a recent Okta survey, 90% of executives said they knew which AI tools their organization used — yet 52% of knowledge workers admitted using unapproved ones. Gucci offers a template: after being slow to see how the internet and influencers changed its market, it formed a shadow board of millennial employees in 2015 that met regularly with senior executives. First step: build an advisory panel of frontline staff and heavy AI users, with a standing quarterly agenda item and a direct line to the board. A panel given an audience but no authority quickly becomes theater.

Stake. Today's AI, infrastructure and talent decisions play out over decades. In 2024, Microsoft signed a 20-year agreement to buy power from a restarted nuclear reactor for its data centers — a commitment stretching into the late 2040s. Directors who left S&P 500 boards this year had served an average of 11.3 years; almost no one in the room for a 20-year deal will be there when it ends. Founders and executives still building careers will live in the world those decisions create. First step: apply a "2040 test" — for decisions with consequences beyond a decade, require a written review from the directors with the longest stake before the vote.

Do boards have to give up experience?

No. The average S&P 500 board has just under 11 seats; allocating one or two to directors who pass the three tests won't dilute it. Nor should those become "AI seats" — AI oversight can't be delegated to one or two directors, since every director has to be able to ask the right questions. What these directors add is a different line of sight that sharpens the whole board's oversight.

The window is closing: either companies decide now who will watch for what the fortifications ignore, or the next threat will come from places they can't see.