As generative artificial intelligence (AI) rapidly penetrates industrial sectors, a widening gap in AI utilization between large and small businesses has emerged. The shift of AI responsibilities from simple tasks to core functions such as software development, supply chain management, and marketing has made the cost burden of AI a pressing issue for companies. AI is evolving into a 'core talent' for businesses, accelerating the restructuring of management frameworks.
According to a report released on August 31 by global consulting firm McKinsey & Company, titled 'The State of AI in 2026: On the Road to ROI,' 40% of companies with annual revenues exceeding $1 billion reported using at least one AI agent. This marks a 13 percentage point increase from last year’s 27%. In contrast, the utilization rate of AI agents among companies with revenues below $1 billion remained steady at 22%.
McKinsey's analysis highlights a clear disparity in the pace of AI adoption between large corporations and small to medium-sized enterprises (SMEs). The stages of AI agent adoption also varied significantly by company size. Among firms with revenues over $1 billion, 40% are in the full-scale deployment phase, followed by 27% not yet adopting, 17% in pilot phases, and 16% in experimentation. For companies with revenues under $1 billion, the highest percentage, 41%, have not adopted AI, while 22% are in full-scale deployment, 19% in experimentation, and 18% in pilot phases. The most widely used tool is AI chatbots at 47%, followed by AI agents and software coding agents, each at around 20%.
Forty-four percent of respondents indicated that AI has entered a full-scale expansion phase, a 6 percentage point increase from last year. However, the perceived impact of AI adoption varies significantly by company size. Among firms with revenues over $1 billion, 54% reported positive effects, exceeding the average by 10 percentage points, while SMEs reported only 33%, which is 11 percentage points below the average.
By industry, the technology sector, including IT, knowledge management, and software engineering, had the highest utilization rate at 31%, followed by consumer goods and retail marketing at 15%, and advanced manufacturing sectors such as automotive, aerospace, and semiconductors at 14%. McKinsey noted, 'In the technology sector, coding; in consumer goods retail, marketing; and in advanced manufacturing, supply chain and production are the battlegrounds for AI deployment,' indicating that AI is being applied first to the most cost-intensive core functions.
AI agents are also reshaping corporate IT budget allocation. A notable trend is the reduction in software purchasing costs through the use of coding agents. Survey results revealed that one in five companies is building software in-house instead of purchasing it. Consequently, 32% reported having never purchased one or more software products. By industry, the highest rates were in technology (41%), healthcare (39%), services (38%), and energy and materials (38%).
Riven van der Beek, a senior partner at McKinsey & Company, stated, 'The rise of software coding agents is fostering a culture of in-house development, shifting the trend among large corporations that previously emphasized external partnerships for AI technology development. AI is transforming companies from consumers in the software market to producers, leading to more cautious IT budget execution.'
The primary challenge for companies remains the cost burden of AI. The survey found that two out of ten companies reported that operational costs, including token fees, are limiting their AI utilization. Recently, AI agents have evolved into a cost structure that increases with model call volumes and data processing.
Despite the cost burden, six out of ten companies indicated they plan to increase AI investments over the next year. Notably, companies that have achieved tangible results from AI—defined as those where AI has impacted EBIT by at least 5%—show a stronger intent to expand investments.
However, Dan Tinkoff, a senior partner, cautioned, 'While 80% of respondents feel improvements from AI, only 37% have seen actual contributions to their EBIT. Companies need to build operational models that fundamentally redesign workflows based on AI.'
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.