Companies everywhere want to figure out how to make their apps and services “smarter” via artificial intelligence (A.I.) and machine learning. But are they hiring lots of A.I. experts to help them achieve these goals?

According to CompTIA’s latest Tech Jobs Report, there isn’t a lot of A.I.-related hiring right now, at least on a state-by-state basis. Check out the chart:

While you might find this visualization surprising, there are some solid reasons why companies aren’t hiring as many A.I. experts as the hype might lead you to believe. First and foremost, hiring specialists for A.I. jobs can prove an expensive endeavor; A.I. developers and machine learning engineers, for example, can easily earn six-figure annual salaries—and that’s before you throw in bonuses, stock options, and other benefits and perks.

Way back in ye olden days of 2017, for example, Tom Eck, CTO of industry platforms at IBM, told those gathered for Markets Media’s Summer Trading event in New York that “top-tier A.I. researchers are getting paid the salaries of NFL quarterbacks.” At other companies, including OpenAI and Google, A.I. researchers have long made millions of dollars. There are no signs these sky-high salaries will dip anytime soon.

While that sort of compensation is only affordable for enterprise-level firms, that’s not the only reason why companies might be reluctant to throw themselves into the hunt for A.I. talent. The simple fact is, it’s still early days for A.I. and machine learning, and many companies aren’t quite sure how to make the company work for them. The idea of using machine learning models to improve customer service or sales outreach is great, but actually integrating those innovations into a tech stack and a product line is something else entirely—and it may take years for companies to reach that point.

In addition, companies that are experimenting with A.I. aren’t necessarily hiring dedicated A.I. talent. With the proliferation of A.I. tools and frameworks, more tech professionals are exploring the possibilities of A.I. and machine learning as part of their broader jobs. Any A.I. tools integrated into their tech stacks are usually built by an outside vendor.

As A.I. becomes more mainstream, we may see a rise in companies needing A.I. experts and researchers, which could unlock a virtuous cycle of more people training for the profession, ultimately driving salary points down. For the time being, though, A.I. will likely remain a highly specialized field.