A widely cited McKinsey report from late 2025 put GenAI adoption among businesses at 58 percent, up from 33 percent the year before. That number got picked up everywhere, usually with some version of "GenAI goes mainstream" as the framing.
The framing isn't wrong. But it flattens something important. Because what's inside that 58 percent varies enormously, and the gap between "we are technically using GenAI" and "GenAI is producing real business value for us" is wide enough to drive a truck through.
What "adoption" actually measures
When researchers measure GenAI adoption, they're typically asking whether a company has used GenAI tools in at least one business function. That's a low bar. A company where one person in marketing uses ChatGPT to draft social posts counts the same as a company that has rebuilt its entire customer service workflow around AI.
The same McKinsey data showed that only about a quarter of companies using GenAI had deployed it across multiple functions. The majority were running narrow experiments or isolated use cases. The adoption curve is real, but the depth of that adoption is still shallow for most businesses.
Where the value is actually concentrating
The businesses showing the strongest results from GenAI share a few things in common. They're using it in workflows where there's high volume, repetitive judgment, and a clear feedback loop. Customer support. Content production. Data summarization. Internal knowledge management.
They've also invested in making sure their teams know how to use the tools well. Not just "here is the login" but actual training on how to write prompts that work, how to review AI output critically, and where the tools are likely to fall short.
The businesses getting the least value are mostly using GenAI as a fancier search engine. They ask it questions, skim the answers, and move on. The tool is technically in use. But there's no real change in how work gets done.
What the 42 percent who haven't adopted are thinking
The survey data on non-adopters is interesting. The most common reasons given are concerns about accuracy, uncertainty about where to start, and lack of a clear use case. Very few cite cost as the primary barrier, which makes sense given how low the entry cost actually is for most tools.
What this tells us is that the adoption gap isn't mainly a resource gap. It's a clarity and confidence gap. Businesses know GenAI is a thing. They're not sure it's a thing for them, or they're not sure where to start, or they tried something once and didn't get great results and moved on.
What this means for your business
If you're in the 58 percent, the question worth asking is whether you're in the shallow end or the deep end. Using a GenAI tool occasionally is not the same as having it change how your business operates. The former is fine. The latter is where the competitive advantage is.
If you're in the 42 percent, the window for getting ahead of your competition is still open but it's not going to stay open much longer. The businesses that figure this out first will have a meaningful head start on the ones that wait until it's obvious.
The 58 percent headline is real. But the more useful number is probably closer to 10 to 15 percent, which is roughly the share of businesses that have actually changed how they work because of GenAI. That's still a lot of room to move.