I worked on a handful of products using Edison, and was speaking to Intel less than a month ago about their Joule line at a conference, where they assured me Jo

I worked on a handful of products using Edison, and was speaking to Intel less than a month ago about their Joule line at a conference, where they assured me Joule was the future. All of these chipsets had (and still have) huge promise, but have been mired in really puzzling and terrible board design issues. You can tell that there are two different groups at Intel, the "Core" and the "Iot". The Edison was super powerful, price competitive, and an honestly wonderful platform to dev on. YOCTO, while a weird decision, was a pretty vanilla Linux flavor and easy to pick up. With all that promise though, the botched the silicon. The 2nd cpu on Edison, the Quark 100mhz one, never actually worked. They were shutoff in firmware from day 1 because of presumed hardware issues. Even worse (and the reason we stopped using Edison), the SPI bus had so much electrical crosstalk on it from not being properly routed or shielded, you couldn't use it at anything over 25hz with a SINGLE bus endpoint. This removed 90% of the real-world uses for the Edison to drive displays, sensor and motor arrayset al. Intel knew it was a problem and consciously decided not to Rev the board to fix it. Gallileo and Joule are both underpowered and incredibly overpriced devices. Today, the raspberry pi 3 is the hobby standard, and in nearly every real world use case, it is orders of magnitude more performant at 10% or less of the cost. Intel IS is trouble, because this is their third botched attempt to enter the world of embedded computing and mobile computing. First was the Atom, which isn't bad, but is too power constrained to compete with ARM. They made some good efforts here, but the cost is higher and perf/watt significantly lower than ARM. Second was their foray into mobile, trying to branch from Atom. Anyone here ever use an Intel powered phone? Well they spent billions on it, never to have a mass market device actually appear. Same problems - while have equivalent performance to ARM, prices were 30-50% higher and performance per watt was significantly worse. Now here we are with attempt 3. With the same issues. Intel fundamentally doesnt know how to design, manufacture or sell embedded chips. It's a completely different market motion, different customers, different constraints, shorter cycles and much much different competitive landscape. AMD isn't going to "beat" Intel. They have fundamentally the same problems. Both AMD and Intel aren't going to go bankrupt, but they are going to continue the slide into much smaller scale manufacture. They are both being eaten by the dozens of ARM vendors, by the FPGA movement, and by public cloud data centers. It's a reduction by a thousand cuts, making it that much more difficult to do anything about it.