As artificial intelligence enters a phase where businesses expect tangible outcomes, companies continue to invest heavily in a technology that has yet to move the bottom line for most.
According to BCG’s AI Radar 2026 survey of 2,400 executives worldwide, the growing enthusiasm for AI is translating into a likely doubling of investments to 1.7% of their revenues.
But returns from AI implementation haven’t, until now, kept pace with the enthusiasm and investment appetite that business leaders have for the technology.
Evidence for this came from a BCG research study of 1,250 firms worldwide last year, which showed that delivering significant bottom-line value has been beyond the reach of the vast majority of these companies. Only 5% of the companies reported achieving AI value at scale. Another 35% reported seeing some returns but admitted they were not moving far enough or fast enough. 60% of these companies have lagged in building critical AI capabilities and reported minimal revenue and cost gains.
The top 5% AI leaders delivered three times greater cost reduction, 1.6 times higher EBIT margins, and 2.7 times the return on invested capital than their peers.
The obvious question is: what did these companies do differently?
First, they pursue a multi-year strategic ambition concentrating AI where it could create the greatest business value.
One of the biggest mistakes companies make is trying to spread AI across dozens of pilots in the hope that some of them will eventually deliver value. Pilots are relatively easy to launch because they avoid the more difficult challenges of data gaps, legacy systems, and cross-functional dependencies. The result is often a collection of disconnected experiments that consume budgets and management attention without improving business performance.
Leaders approach AI as board- and CEO-sponsored initiatives, ensuring they don’t end up being isolated experiments. They concentrate investment on a handful of business domains where AI could have the biggest impact, nearly 70% of AI’s value potential lies in these core functions research and development, sales & marketing, manufacturing, procurement, and supply chain planning.
Nestlé illustrates this approach well. The world’s biggest food and beverage company set out to reimagine procurement as a strategic lever for growth using AI-enabled simplification to unlock the capital needed to reinvest in AI, innovation and brand-building.
While the company aimed to reduce costs by CHF 2.5 billion ($3.15 billion) by 2027, it generated close to CHF 700 million ($900 million) of that in the first year itself, with the annual savings rate increasing from 3% to 5%. What makes this more than a cost story is the ambition behind it: every Swiss franc saved is being reinvested into innovation, brand-building, and long-term growth.
The Nestlé AI-enabled initiative is an example of what’s possible when strategy, execution, and culture come together at scale.
Second they used AI to reshape and reinvent businesses, not simply automate existing work.
Many organisations view AI primarily as a productivity tool. But the leaders look beyond automation to reshape and reinvent businesses. Nearly 90% of these companies expect most of their AI value to come from reinventing business processes rather than making existing ones more efficient.
A global beauty products company implemented a change in its sales and marketing as it sought to address multiple issues with customers’ online experience. These problem areas included highly fragmented sources of information, a lack of personalized guidance, and an inability to access important product information before making a purchase decision. The company introduced the industry’s first large-scale virtual beauty assistant, integrated into the company’s websites and e-commerce journeys extending across more than 20 markets and covering eight brands. The virtual assistant was deployed in less than 12 months, and the company expects to gain $100 million in incremental revenue from the initiative—doubling the ROI of traditional e-commerce customer pathways.
Third, they recognise that AI transformation is fundamentally about people and organisations.
AI leaders assign their best people to AI projects, signalling the initiative’s importance in the scheme of things. They are also more likely to explicitly define how behaviours need to change, and introduce incentives for the same.
Further, they invest adequately in helping people to adapt to new roles, as AI reshapes work. An effective AI operating model does not focus on replacing people with technology; it entails reimagining the company around AI.
Additionally, future-built firms invest in broad-based employee enablement. They pair bold training ambitions with widespread engagement.
Fourth, they use fit-for-purpose technology and data.
Future-ready companies are aware that capturing value from AI requires more than just adding new tools; it requires a rewiring of the firm’s architecture and processes around the new technology.
Future-built firms are three times as likely to operate a central, integrated AI platform as the backbone for their AI deployment. They reinforce their data foundation with strong governance and are three times as likely to enforce enterprise-wide data policies through central oversight teams, ensuring quality, trust, and responsible use.
To summarise, AI transformation is inherently harder than previous technology transformations.
It is important to know that organisations where CEOs set clear priorities, hold leaders accountable, track value, and devote sufficient focus to change management turn early AI momentum into a lasting advantage.
Time is short. The technology is advancing fast, making catching up more difficult with each passing week. Future-built companies are pulling away, widening the value gap, and putting slow movers in a deeper value hole. The others need to move now.
(The author is Managing Director & Senior Partner, BCG. Views are personal.)