From AI to Robotics: Why Only Some Firms Turn Digital Technologies into Growth

An OECD study finds that while AI and advanced digital technologies are spreading across firms, adoption remains limited and is concentrated among large, skill-rich and already digitalised companies. Productivity gains come mainly from complementary investments in skills and digital capabilities, not from technology adoption alone.

Researchers from the OECD Directorate for Science, Technology and Innovation, working with national statistical offices, universities and policy research institutes across 15 OECD countries, have taken a close look at how artificial intelligence and other advanced digital technologies are actually spreading through the business world. Using detailed firm-level data from countries including Germany, France, Japan, Korea and the United Kingdom, their study cuts through the hype to show where digital transformation is really happening, and where it is not.

The findings challenge the popular idea that AI is already everywhere. Between 2017 and 2023, only a small share of firms, typically between 4% and 10%, reported using AI or 3D printing. Technologies such as big data analysis and the Internet of Things were more common, but still used by only about one in four firms. Despite rapid progress and growing attention, advanced digital technologies remain far from universal in everyday business operations.

Advanced technologies don’t arrive alone

One of the strongest messages from the study is that firms do not adopt AI in isolation. Companies that use AI, robotics or big data analysis almost always already rely on more basic digital tools. Cloud computing, enterprise resource planning software, customer relationship management systems and fast broadband connections form the foundation on which advanced technologies are built.

In simple terms, firms tend to climb a digital ladder. They start with basic systems that organise data and processes, then move gradually toward more advanced applications. This explains why digital transformation often looks slow from the outside. It is less about sudden breakthroughs and more about the steady accumulation of capabilities over time.

Why some sectors race ahead

The spread of advanced digital technologies varies sharply by sector, and these patterns are remarkably similar across countries. AI and big data analysis are most common in ICT and professional and scientific services, where data, analytics and highly skilled workers already play a central role. Manufacturing and utilities lead in robotics and 3D printing, reflecting their importance for industrial automation and production.

The Internet of Things stands out as more versatile. It is widely used not only in ICT, but also in construction, transport, logistics and manufacturing, where sensors and connected devices help monitor equipment and manage physical processes. These differences show that technology adoption is closely tied to how firms actually operate, rather than following a single universal path.

Size and skills still matter most

Firm size turns out to be one of the strongest predictors of adoption. Larger firms are far more likely to use AI and other advanced technologies than small and medium-sized enterprises, even when they operate in the same sectors. In several countries, large firms are around 20 percentage points more likely to adopt AI than are otherwise similar small firms.

Skills matter just as much. Firms that employ ICT specialists, invest in digital training and already use a wide range of digital tools are much more likely to adopt advanced technologies. Where detailed worker data are available, higher shares of university-educated employees and technical occupations strongly increase the chances of adoption. Technology, the study shows, does not replace skills, it depends on them.

Productivity gains: promise with conditions

At first glance, firms that adopt AI, big data analysis or robotics appear more productive than those that do not. In simple comparisons, adopters often show higher output per worker, sometimes by large margins. Big data analysis and robotics, in particular, are consistently linked with higher productivity.

But the picture changes once skills, training and digital readiness are taken into account. When researchers control for these factors, most of the apparent productivity advantage of AI disappears. This suggests that it is not AI alone that drives performance, but the broader capabilities of the firms that adopt it. Highly digital, skill-rich firms both adopt earlier and perform better, creating the impression that technology itself is doing the work.

The study’s message is clear and timely. AI and advanced digital technologies have enormous potential, but their economic benefits depend on foundations that take time to build. Without investment in skills, digital infrastructure and organisational change, technology alone is unlikely to deliver the productivity boost many expect.