One Role Just Became the Shortest Path to CEO

For decades the route to the top job was a known quantity. Run operations, become the COO, wait your turn. Boards liked it because it was safe. The heir apparent had seen the whole machine, knew where the bodies were buried, and could keep the trains running from day one.

That machine is breaking, and the data is blunt about it.

In 2018, COOs accounted for 41.8% of insider CEO appointments across the Russell 3000. By 2025 that figure had fallen to 21.1%. Half the pipeline, gone in seven years. And in 2025, for the first time in at least five years, S&P 500 boards chose outside CEOs more often than the sitting COO.

Boards have not stopped doing succession planning. They have stopped believing the old plan produces the right leader. The question is what changed, and I think the answer is sitting in plain sight.

Execution got cheap

In 2025 a developer named Maor Shlomo built Base44 entirely alone. No employees, no funding round. Six months after launch, Wix bought it for 80 million dollars. He is not a one-off. Roughly 36% of new ventures in 2026 are solo-founded, up from under a quarter in 2019.

Look closely at what these founders are actually doing. They are not doing everything. They are deciding everything, and directing a stack of AI agents and tools that do the rest. Code, design, support, content, outbound. The functions that used to require a team now require a decision.

This is the part I think most commentary misses. When execution collapses in cost, the binding constraint on a company moves. It is no longer "can we build it and sell it." It is "which problem, which customer, which sequence, and what do we say no to." The scarce skill shifts from doing to deciding.

Now scale that up. A CEO has always been an orchestrator of functions they could not personally perform. The difference is that orchestrating those functions used to mean managing thousands of people through layers of executives. Increasingly it means directing systems, with far fewer humans in between. The CEO job is being compressed toward pure judgment.

So which discipline trains for exactly that?

Deciding is the product craft

Product is the only function whose entire job is making calls across every other function without the authority to force any of them. A good product leader spends their career choosing problems, killing loved ideas, sequencing bets, and aligning engineering, design, marketing and sales around a decision they cannot simply mandate. Influence through judgment, not headcount.

That used to be product's weakness. No army, no budget of its own, accountability without control. In a world where the armies are dissolving into tooling, it is starting to look like the best CEO apprenticeship available.

Boards are noticing. In 2022, about 15% of Fortune 1000 companies had a Chief Product Officer. By 2025 it was closing on 60%, and around 70% of those CPOs hold P&L responsibility. The role is already running businesses, not backlogs. One industry study goes further and predicts that by 2030, 30% of CEOs will be former CPOs.

A caveat I would be dishonest to skip: that last prediction comes from Products That Count, a community of product people forecasting the rise of product people. Take the number with salt. The succession data from the Conference Board carries no such conflict, and it points the same direction.

The obvious objection

Here is the pushback I expect, because it is a good one. If AI lets product people absorb engineering, it equally lets engineers absorb product. Why does the PM's skill survive while everyone else's gets commoditised?

The honest answer is that the PM title probably does not survive, at least not in its current form. LinkedIn has already replaced its Associate Product Manager program with a Product Builder track that spans product, design and engineering. The role is dissolving into the team, and plenty of smart people predict full-time PMs disappear entirely.

But watch what is dissolving and what is not. The parts of product work that AI absorbs are the artifacts: the PRDs, the tickets, the research summaries, the coordination overhead. The part it cannot absorb is the judgment about customers and markets, because that judgment is not verifiable by a machine the way code is. An AI can tell you whether software works. It cannot tell you whether it matters.

So the title dissolves downward while the judgment gets promoted upward. Everyone can build. Someone still has to decide what is worth building, and that person increasingly runs the company. Whether their business card ever said "product manager" matters less than whether they spent years practising the craft.

The same logic is eating the last refuge of the sales-led CEO argument, by the way. Distribution used to be a career-length moat of relationships and grind. The rise of GTM engineering is turning most of it into systems design: enrichment, intent signals, automated outbound, personalisation at scale. Worth being straight about this too. The people winning at GTM engineering today mostly come from RevOps and sales ops, not product. The moat is not falling to product people specifically. It is just falling, which means no discipline owns distribution anymore, and the generalist who can design the system and understand the customer is well placed to pick it up.

What I would do with this

If you lead product and any of this rings true, the move is not to wait for boards to figure it out. It is to make yourself legible as a CEO candidate now.

Take the P&L. If your role does not carry revenue and cost ownership, ask for it before you ask for anything else. Boards do not read roadmaps as CEO readiness. They read P&Ls.

Own one distribution system end to end. Not the strategy deck, the actual loop. Build or run one automated acquisition system yourself. Distribution fluency is the gap most product leaders have, and it is the cheapest it has ever been to close.

Practise orchestration, not management. Run a meaningful slice of your function on agents this quarter and be able to talk concretely about what you automated, what you kept human, and why. Orchestrating hybrid teams of people and systems is the actual job description of the next decade's CEO, and almost nobody can demonstrate it yet.

The succession machine that reliably produced operations-shaped CEOs is winding down. What replaces it will select for judgment under ambiguity, customer instinct, and the ability to run a company as a system rather than a headcount. Product people have been training for that job for twenty years without knowing it.

The seat is opening. Walk toward it.

Sources: The Conference Board / Semler Brossy, CEO Succession Practices 2025. BCG, From COO to CEO, 2026. Forbes, April 2026. Products That Count, CPO Insights Report. Itamar Gilad, Will AI Eliminate PM Jobs?, 2026.

Barry Freeman is a product executive and founder of SparkQuest, a Melbourne AI product studio.