I think part of the confusion is that Apple reports units shipped into the channel, and the analysts are talking about units sold.
[Edit: Strod, given 20% year-over-year in the US, looks like reality was smack in the middle of the two predictions.]
Ok, I came back to this article to see who was closer, but now I'm just confused, because Apple's Macs-shipped numbers for the Americas, according to their own PDF, show 971K CPUs, which is a 20% increase from the same quarter last year. Gartner said 34% growth, so they missed that, and IDC said 8.3%, which Apple beat comfortably.
Yet the table in this article shows 971K computers as being a HUGE decrease versus last year, and that 971K is for the entire Americas, not just the US.
It also shows 1.398M Q1 2010 shipments, a number I don't see anywhere on Apple's real numbers--the actual units shipped to the Americas was 1.187M in that quarter, and worldwide was 3.362M.
Where the hell are these research firms getting these numbers from? Somebody better informed care to explain?
(Also, unrelated, if you count iPhones and iPads as CPUs, which you definitely should for iPads and probably should for iPhones and iPod Touches as well, Apple's numbers look a LOT better. Given that that's where the future of average-consumer computing is headed, 10% marketshare is paltry. I don't think that means they'll stop selling Macs, just that they're going to end up being "pro" machines for developers and such, which is a small but profitable market, like the XServe is now.)