Corporate Metabolism

Salman Ahmad

Demis Hassabis, who runs Google DeepMind and won the Nobel Prize for the AlphaFold breakthrough, recently described what is coming as "10 times the impact of the Industrial Revolution, at 10 times the speed". The Industrial Revolution unfolded over a century. The transformation enabled by AI is compressing into a decade. That is not a forecast about technology. It is a forecast about every enterprise, every workforce, and every operating model.

Andy Grove had described moments like this as strategic inflection points in Only the Paranoid Survive: shifts so fundamental that the rules of competition change while incumbents are still operating under the old ones. AI is one of those moments. And most enterprises are not built for it. McKinsey's latest State of AI report captures the gap in a single contrast: 88 percent of organizations at the turn of the year used AI in at least one business function, yet only about one-third reported scaling it across the enterprise, and roughly 6 percent qualified as high performers capturing meaningful financial impact. The technology has scaled. The enterprise response has not.

The metabolism most companies were built for no longer matches the environment

Large companies were designed for episodic change. A major technology shift would arrive every several years, the company would absorb it through a structured program, and execution would unfold over eighteen to thirty-six months. The model worked because the underlying capability was stable once chosen. There was time to plan, align, build, and scale. There was time for the system to digest.

What we have learnt at the Executive Technology Board is that AI breaks that pattern. The capability is not stable. It shifts under the program while the program is still being executed. A roadmap written in January is partly obsolete by July. The governance model assumes the world holds still long enough for a decision to be made and implemented. It no longer does. In an environment where the underlying capability resets every quarter, slow decision cycles produce something worse than delay. By the time a decision is made and built, the thing being built is often no longer the right thing.

The cautious-feeling choice is rarely the safer one

Inflection points punish hesitation. Waiting feels like the conservative move, which is exactly why most large enterprises default to it. But in a moment when the underlying capability is shifting every quarter and competitors are reshaping their operating models in real time, waiting is not caution. It is exposure. The companies that move late tend to stay late, because the advantages built by early movers shape the rules of competition that follow.

Boards and CEOs intuit this. The challenge is that almost everything inside a large enterprise pushes the other way. Governance is built to slow things down. Risk frameworks are biased toward avoiding loss rather than capturing upside. Annual planning rewards predictability over responsiveness. The hardest leadership work right now is recognizing that the default behavior of a large enterprise is to move too slowly through this moment. The structure favors waiting. The leader's job is to actively counteract that pull.

Funding, decision rights, and governance all need to operate at a different pace

The issue is not ambition. Most CEOs and boards I work with are ambitious about AI. The issue is organizational design. Who can approve experimentation. How quickly do resources move. How fast learning from one deployment shapes the next. These are mechanical questions, and the answers in most large enterprises are wrong for the current moment.

Three areas need to shift. Funding needs to become more dynamic, with smaller and faster mechanisms that reallocate capital as evidence emerges, rather than annual cycles that lock decisions in for twelve months at a time. Decision rights need to move closer to the work, with clearer guardrails defined in advance, so business leaders can act without re-litigating each step. And governance itself needs to shift from gate-keeping to monitoring, because AI outputs drift and behavior changes in production. Risk cannot be assessed once and then managed. It has to be evaluated continuously.

Jeff Bezos's distinction between Type 1 and Type 2 decisions, from his shareholder letter now a decade ago, captures an important operating principle underneath all of this. Type 1 decisions are consequential and irreversible - the one-way doors that demand careful deliberation. Type 2 decisions are reversible and can be unmade if they turn out wrong - these are two-way doors. Bezos observed that as organizations get larger, they tend to use the heavyweight Type 1 process on most decisions, including many that are clearly Type 2, producing slowness and unthoughtful risk aversion. That is exactly what is happening with AI in most enterprises today, where decisions that are genuinely Type 2 get routed through Type 1 governance. The shift is to identify which decisions go through a one-way door and govern those carefully, and let the rest move quicker through two-way doors.

Innovation at the pace of AI is a management challenge, not a technology challenge

A healthier metabolism has a few consistent traits. Decision loops are shorter, measured in weeks rather than quarters. Business leaders become owners of workflow redesign, adoption, and value creation, not passive sponsors of technology projects. Risk and legal are embedded in the work rather than consulted after it. Pre-cleared patterns get reused, so each new use case does not re-open every prior question. Real evaluation infrastructure catches issues continuously, rather than annual audits catching them late.

None of this is necessarily new. The patterns exist, and they are visible in smaller AI-native companies I work with and in the pockets of large enterprises that have been deliberately structured to move faster - but often as carve-outs or new business builds. What is missing in most cases is the conviction to redesign the broader operating model around these patterns, not just to graft a few practices onto an unchanged structure.

The companies pulling ahead are not just adopting AI faster. They are redesigning themselves so the enterprise can learn, decide, and adapt faster without losing control. The technology choices matter, but they are downstream. The upstream choice is the one boards and CEOs are now being asked to make: how the company decides, funds, and governs at the pace its environment now demands.

Are you intentional about the one-way vs two-way doors in your organization - and does that clarity permeate the entire organization? My sense is the corporate metabolism gap, at the moment, needs real attention. And the window to address it is shorter than most boards assume.