.jpg)
In recent years, so many conversations about AI transformations in the corporate world have seemed to go the way that homeowners talk about remodels: An upgrade would be nice and would add a lot of value to the place; let’s do it sometime this year; let’s move forward once the plans are finalized, the budget’s approved. Then, things stall. Months tick by, and then a year.
Meanwhile, prices are increasing, and other problems arise—maybe even ones that the upgrade you have been talking about could’ve helped avoid. The reality hits: You may think you can’t afford to make the changes, but the truth is, you can’t afford not to.
The new Build for the Future 2025 report from BCG found that a small cohort—just 5 percent of companies—are generating substantial value from AI. But that value is twice the revenue increase and 40 percent greater cost reductions when compared to companies that haven’t invested in AI capabilities. This 5 percent has figured out how to extract real, scaled value from AI, and that value is going to continue to grow, as is the way of technological gains. Meanwhile, 60 percent of the more than 1,200 companies studied for the report have little or nothing concrete to show for their investments. The degree of separation between early adopters and those who haven’t made moves to integrate AI thoughtfully isn’t a slight gap anymore; it’s a widening chasm. That’s because, in practice, many companies confuse “being cautious with “moving slowly.”
“Leaders tend to distance themselves from the rest of the pack more and more,” says Nicolas de Bellefonds, global leader of AI at BCG and a co-author of the report. “The time for passive observation has sailed.” That’s not just advice to integrate AI, it’s also a result of the study itself. AI technology is advancing fast enough that every week spent observing instead of acting makes catching up harder. Future-built companies—BCG’s term for that 5 percent—are “pulling away, widening the value gap, and putting slow movers in a deeper value hole.”
To understand how future-built companies got that way, the report looks at how they applied AI technology and how they spend their money on it. It finds that the biggest value isn’t in side projects or one-off chatbots—it’s in the heart of the company: sales, manufacturing, supply chain, R&D, and, increasingly, IT. “Big value comes not from AI pilots or isolated use cases,” the report notes, “but from reshaping and reinventing core business workflows end-to-end.”
Amanda Luther, Global Leader of AI and Digital Transformation at BCG and a report co-author, says that “focusing on a few big bets” that you transform across the value chain can have the biggest return on investment. “If you’re a consumer-goods company, pick marketing or R&D and go for material lift. If you’re industrial, it might be manufacturing and supply chain,” she says. “It's about figuring out what those couple of places are that really matter for the competitive advantage of the company, and then really driving against those.”
The report doesn’t serve as a post-mortem on how certain companies have become AI leaders, deriving value from the technology in their teams and in their businesses, but as a proof of concept for fast followers. It’s not about catching up as much as it is about making a move now and having data to back it up.
“I think the positive news for those that hadn't moved yet is that there is proof at scale at other organizations of where value delivery can really happen,” explains Luther. “It can allow you as an organization to be a fast follower and to focus on those areas where there's real value. But I do think there is a widening gap. It's largely because the companies that have already started have fixed their data foundations. They've started to create an organization that is willing and able to use AI. There's kind of an organizational muscle for adopting technology and change that the early adopters and those leaders have built that will continue to get stronger.”
De Bellefonds agrees, and that catching up isn’t a competition for the “highest number of use cases.” That’s in part because leaders are seeing the value of the technology and reinvesting their returns. Those included in the study plan to spend twice as much on IT this year and direct more of that IT budget specifically to AI.
The impact of agentic AI is one of the fastest accelerators. What once seemed like a buzzword is now proving its mettle, as the report estimates agentic AI—systems that reason, learn, and act autonomously across workflows—account for 17 percent of AI value in 2025 and could reach 29 percent by 2028. To achieve that value, the research shows that deeply integrating agentic AI into the area that’s been identified as most important to the company—like the marketing or supply chain areas that Luther referenced—is best practice.
Like any transformation that has a significant ROI, it takes time. And hard work.
“There’s this misconception that AI is this magic wand and integrating it will be easy—that you just have to plug this tool and voila! it all works out,” says de Bellefonds. “That misconception, frankly, is often pushed by some AI providers. The fact is, it takes a bit of hard work to really transform how you work and how you operate. People underestimate that. Maybe because they want to believe that it's easy. The report shows there is real value, but it only works if you do the real transformative work that goes with it.”
That means AI failure is not often about the model being used. The report found that most roadblocks are human and organizational, like aligning strategies, training, getting people to actually use the tools, handling unstructured data, and setting clear goals. Companies that skip the groundwork—operating model, skills, guardrails—struggle to scale, no matter how clear their prompts or clean their coding.
“We always had this rule of thumb we called 10, 20, 70,” Luther says. The rule suggests that success with AI relies 10 percent on the algorithms, 20 percent on the tech, and 70 percent on the people. With this longitudinal study, “it actually proves out… that it is really all about the people.” Done well, she adds, AI “takes the toil out of the day-to-day” and “creates more joy in the job” because humans spend more time on judgment and creativity.
The performance case for AI, it turns out, is also a human case.
