Days before Apple’s WWDC event, Microsoft’s own AI laptop initiative just got some bad news. While the first Copilot+ devices are set to launch on June 18, it turns out only a portion of new laptops being sold under the Copilot+ name are going to actually have any kind of advanced AI features built-in at launch.

This comes as a bit of a surprise, as processor companies AMD and Intel both announced at Computex that they would be releasing Copilot+ compatible laptop chips later this fall. Prior to that, Microsoft had only ever talked about Copilot+ coming to Windows on Arm devices, which use chips from Qualcomm.

That’s a lot of branding to throw at you, but what this essentially means is that Copilot+ laptops are now set to be split between three different chipsets—and unfortunately, it looks like two of those chipsets are going to lag behind the other.

In statements to The Verge, representatives from Microsoft, Intel, AMD, and Nvidia (which is working with AMD) confirmed the non-Qualcomm laptops are going to have to wait for their Copilot+ features. While Arm-based laptops will arrive later this month with access to Recall and the gaming-focused Auto Super Resolution mode, laptops with AMD and Intel chips will have to wait for “free updates” to get those features, Microsoft marketing manager James Howell told The Verge.

More specifically, AMD PR manager Matthew Hurwitz indicated these “Copilot+ experiences” might not arrive until “the end of 2024.”

Perhaps some of this was to be expected. Intel’s own press release for its AI laptop chips mentions that Copilot+ features will come via free updates “when available.” But language and potential timetables weren’t entirely clear until now.

While Microsoft does not directly manufacture AMD or Intel devices, with the exception of certain Surface models, this does put the Windows maker in a tricky spot as Apple prepares its own AI features to be rolled into macOS 15. Generally, an Apple user will be able to buy a new Mac and assume it can do all Mac things; Windows users looking to get into AI will be split across three options, and will have to do a lot more research about what they can and can’t do with their new machines right out of the box.