Reading time
Series: Out-innovate your competition with your operating model — Post 1 of 5
BCG scored more than 1,250 companies worldwide on AI maturity across 41 dimensions in its September 2025 study, The Widening AI Value Gap. Only 5% are creating value from AI at scale. Another 35% are scaling and starting to see returns, and the remaining 60% are either experimenting without results or have made no real move at all.
The obvious explanation would be that the leaders simply bought better technology. BCG’s own numbers point elsewhere. Future-built companies plan to spend 26% more on IT overall and to put up to 64% more of that budget into AI, which works out to roughly twice the AI investment of the laggards. The difference is that they have somewhere to put the money that pays it back, so they keep spending.
The performance gap is not a rounding error either. Future-built companies show 1.7 times the revenue growth, 1.6 times the EBIT margin and 3.6 times the three-year shareholder return of the laggards, and that last figure comes from market data rather than from anything the respondents said about themselves. BCG measures the two alongside each other rather than proving that one caused the other. Either way, no vendor sells that gap.
I get a close-up look at that gap more often than I would like. A company buys Copilot for a few hundred people and tells itself it is now an AI company. Six months later most of those people still are not sure what they are supposed to do with it. Nobody set a target, nobody recorded how the work performed before the tool arrived, and nobody took on the unglamorous job of changing how people actually work. The licenses keep billing every month.
Building has never been faster or cheaper. An idea becomes a working prototype in an afternoon, and teams are producing more code, more decks and more experiments than at any point I can remember. Most of it never reaches a customer.
S&P Global Market Intelligence asked more than 1,000 IT and business professionals in North America and Europe to estimate how many of their proofs of concept never reach production. The average answer was 46%. Over the same year, the share of respondents reporting that their organization had abandoned most of its AI work went from 17% to 42%. These are estimates from the inside rather than an audit, and a jump of that size probably includes people counting differently than they did a year earlier. It still points one way.
Everyone got busier, and the numbers in front of the board stayed where they were. If you have felt that gap inside your own organization, where a wall of AI activity sits next to a strangely quiet set of results, you are not imagining it.
The natural reaction is to doubt the technology or believe you supported the incorrect provider when the value doesn’t materialize. That is rarely where the issue is hidden, in my experience. Give a thousand people a license, and you’ll have a thousand employees who work a little faster at their own desks within a company that operates precisely as it did a month before but now has a larger bill.
The sector data points the same way. Software, telecommunications and payments and fintech lead BCG’s 2025 maturity ranking, while fashion and luxury, chemicals, and real estate and construction sit at the bottom. BCG does not say why. My own reading, at least for software and fintech, is that both already had a short path from a working idea to a paying customer before AI arrived. AI paid out on a machine that was already running.
BCG also measured what separates the two groups organizationally, and one number stands out. Nearly 100% of future-built companies report a deeply engaged C-suite, against 8% of the laggards. They are three times more likely to have appointed a chief AI officer and two and a half times more likely to have governance or a way of measuring AI value already in place. None of that arrives with a license.
Describe the conditions that must be met for an AI project to be sustained. It must be funded before ten other proposals vie for the same funding. The messy task of transforming a team’s operations must be taken on by someone; this is a whole separate task from implementing software. To demonstrate what changed six months later instead of debating it over a slide, someone should preferably have documented the beginning point.
The technology doesn’t contain any of that. It resides in the organization’s operating model, which includes how it chooses what to support, how it oversees the work, who promotes adoption, and whether anyone cares to keep track. Accelerating things won’t save that machinery if it wasn’t able to convert effort into value prior to the advent of AI. Simply said, you will produce the unused work more quickly. A corporation has never outperformed its competitors by producing more of something that no one uses, and AI does not alter that math.
That’s precisely why AI proves to be such a harsh mirror. The malfunction was not caused by it. It just made it impossible to ignore, and it did so quickly.
The businesses making significant profits from AI are not operating on a proprietary model that is unavailable to the public. They have established an organization capable of absorbing the output of technology. BCG lists five strategies that its leaders follow, starting with leadership from the top and running through workflows, operating model, talent, and technology and data. Four of the five sit in the organization rather than in the stack. When I simplify that down to my own shorthand, it comes down to three questions that most leadership teams have never satisfactorily addressed.
Funding
How do you choose which AI projects receive funding, and—the more difficult part—how do you eliminate the ones that don’t work? Most AI expenditures are distributed across a dozen disconnected pilots, none of which have a clear owner or a stop point. The entire situation changes if it is conducted more like a portfolio than a science fair.
Adoption
Who is responsible for the change in the way the work is completed? Not the initial email. The real actions. If no one owns that shift, you have purchased shelfware with a subscription attached. A license is not a capacity.
Measurement
Has anyone recorded a baseline? Almost everyone ignores this phase, which silently destroys the business case. Without a read of the work’s performance before the tool’s arrival, ROI becomes a matter of opinion and you have nothing to compare it to later. Opinion always loses out to numbers in budget reviews.
When those three are right, AI becomes a real advantage. Get them wrong and BCG’s own numbers put you in the 60% that is still waiting. The report is blunt about who the playbook is for, calling it a roadmap that the other 95% can use.
This is not an argument against artificial intelligence. It is a justification for creating an operational model that enables you to profit from it, ideally before your rivals do.
I’ll get specific in the next four articles. How to manage and finance AI projects at the portfolio level. Why regulated sectors, such as pharmaceuticals and clinical research, are not the outlier that many believe they are. The budget-draining tooling trap that goes unnoticed. And what happens over the first ninety days of developing a value-focused operational strategy.
One question to consider for your upcoming leadership meeting for the time being. How much of everything your teams have created this year with AI is actually being used, and could you provide evidence if someone asked?
You are not behind the pack if your sincere response is that you’re unsure. Most businesses are also unable to respond. The potential is right there—you haven’t yet transformed this into an operating model question.
Artikel delen
For Product Owners, gaining a certification has become standard practice in many organisations. Scrum, SAFe and other Agile frameworks are
By Nico Schellingerhout and Jeroen Jan Elzinga Agile. A word that once promised transformation, but is now often met with
A summary of the webinar: Strategic portfolio management: what organisations can learn from Silicon Valley’s VCs, 17 June 2026. Why
Change does not have to be difficult or complex. With the right focus, energy and guidance, you create clarity, direction and results.
We would be happy to explore how we can help.
