By Jacques Bughin and Nicolas van Zeebroeck of McKinsey The diffusion of a new technology, whether ATMs in banking or radio-frequency identification tags in retailing, typically traces an S-curve. Early on, a few power users bet heavily on the innovation. Then, over time, as more companies rush to embrace the technology and capture the potential gains, the market […] 
By Jacques Bughin and Nicolas van Zeebroeck of McKinsey
The diffusion of a new technology, whether ATMs in banking or radio-frequency identification tags in retailing, typically traces an S-curve. Early on, a few power users bet heavily on the innovation. Then, over time, as more companies rush to embrace the technology and capture the potential gains, the market opportunities for non-adopters dwindle. The cycle draws to a close with slow movers suffering damage.
Our research suggests that a technology race has started along the S-curve for artificial intelligence (AI), a set of new technologies now in the early stages of deployment. It appears that AI adopters can’t flourish without a solid base of core and advanced digital technologies. Companies that can assemble this bundle of capabilities are starting to pull away from the pack and will probably be AI’s ultimate winners.
Executives are becoming aware of what is at stake: our survey research shows that 45 percent of executives who have yet to invest in AI fear falling behind competitively. Our statistical analysis suggests that faced with AI-fueled competitive threats, companies are twice as likely to embrace AI as they were to adopt new technologies in past technology cycles.
AI builds on other technologies
To date, though, only a fraction of companies—about 10 percent—have tried to diffuse AI across the enterprise, and less than half of those companies are power users, diffusing a majority of the ten fundamental AI technologies. An additional quarter of companies have tested AI to a limited extent, while a long tail of two-thirds of companies have yet to adopt any AI technologies at all.
The adoption of AI, we found, is part of a continuum, the latest stage of investment beyond core and advanced digital technologies. To understand the relationship between a company’s digital capabilities and its ability to deploy the new tools, we looked at the specific technologies at the heart of AI. Our model tested the extent to which underlying clusters of core digital technologies (cloud computing, mobile, and the web) and of more advanced technologies (big data and advanced analytics) affected the likelihood that a company would adopt AI. As Exhibit 1 shows, companies with a strong base in these core areas were statistically more likely to have adopted each of the AI tools—about 30 percent more likely when the two clusters of technologies are combined. These companies presumably were better able to integrate AI with existing digital technologies, and that gave them a head start. This result is in keeping with what we have learned from our survey work. Seventy-five percent of the companies that adopted AI depended on knowledge gained from applying and mastering existing digital capabilities to do so.
This digital substructure is still lacking in many companies, and that may be slowing the diffusion of AI. We estimate that only one in three companies had fully diffused the underlying digital technologies and that the biggest gaps were in more recent tools, such as big data, analytics, and the cloud. This weak base, according to our estimates, has put AI out of reach for a fifth of the companies we studied.
Leaders and laggards
Beyond the capability gap, there’s another explanation for the slower adoption of AI among some companies: they may believe that the case for it remains unproved or that it is a moving target and that advances in the offing will give them the chance to leapfrog to leadership positions without a need for early investments.