Since AI went mainstream a few short years ago, most companies have run small AI experiments. But while chatbots and recommendation engines proved to be interesting add-ons, they were rarely business-critical. In 2025, 88% of organizations are regularly using AI, up significantly from 78% in 2024. And for some industry leaders, AI has gone from an innovation project to an operational requirement.
From Early Adopters to High Performers
64% of respondents to McKinsey’s latest global survey on AI say that AI technology is enabling their innovation. Marketing and sales departments were early adopters, using AI for content creation and customer targeting. IT departments followed, deploying AI for service desk management and knowledge systems. High performers use AI for efficiency but also for growth and innovation.
Manufacturing has embraced AI for predictive maintenance and quality control, while healthcare organizations use it to analyze medical images and assist with diagnoses. Financial services firms deploy it for fraud detection and risk assessment, and energy companies are exploring AI applications for optimizing operations and reducing carbon footprints.
From Pilots to Enterprise Scale
Building an AI proof-of-concept was never the challenge. Getting thousands of employees to actually use it every day is where most organizations stall. Only 39% report that AI has affected their enterprise-level earnings. The rest are stuck running successful tests that never expand beyond a single team.
The organizations that are scaling most successfully are rebuilding their core processes around what AI can do. This requires executive commitment, role-based training programs, and mechanisms to incorporate feedback and constantly improve AI systems. In fact, half of all high performers intend to use AI to transform their businesses, not just improve them.
“Building an AI pilot is easy; getting it used every day across an organization is the hard part,” says Ivo Bozukov. “In my experience, scaling only happens when leadership commits to changing processes, training people properly, and treating AI as a long-term capability rather than a short-term project.”
The Rise of Agentic AI
Early generative AI tools were impressive but very limited. The most basic use involved asking a question and getting an answer. Newer systems are agentic, meaning they can plan and execute multiple steps in a workflow without constant human supervision.
Twenty-three percent of organizations are now scaling agentic AI somewhere in their operations, with another 39% experimenting with AI agents. IT and knowledge management functions lead agentic AI adoption, with use cases like service desk management and deep research. For business leaders like Ivaylo Bozoukov, understanding these technical shifts is essential to strategic planning in technology-dependent industries.
From Adoption to Impact
While AI is being adopted at record speed, clear, measurable results remain rare. Those organizations that are turning investment into impact are treating AI as a transformational tool that can optimize every aspect of their operations.
But scaling AI also requires massive physical infrastructure. The first ten Stargate facilities represent hundreds of billions in investment and gigawatts of power capacity. Ivo Bozukov has seen that without that infrastructure, the technology can’t scale, no matter how advanced AI technology becomes.
Ultimately, scaling AI means redesigning how organizations work and building the infrastructure to power it. The organizations that successfully overcome these challenges are positioned to lead their industries, while those that fail risk falling behind.
