Intel has confirmed in the Q4 earnings report that Skylake will land in the second half of 2015 and that 10nm is still far off.
Intel's CEO Brian Krzanich confirmed that the Broadwell delay has not affected the Skylake launch schedule. The company still expects to release Skylake in the second half of the year in the back-to-school time frame. The CEO also stated that they're still timing the 10nm manufacturing process. Intel expressed the intent to reveal the progress with 10nm and its approximate time of introduction into the market at the November conference call.
Intel Confirms Skylake in 2H 2015 - 10nm Still Far Off
According to Intel's tick-tock cadence Broadwell was supposed to land in the first half of 2014. However, due to various delays mandated by a troublesome 14nm process, Broadwell parts, especially the high performance variants, were pushed back to 2015. The company managed to squeeze in the tiny core-M parts in late 2014 and the slightly larger Broadwell-U parts with the HD 6000 graphics in early 2015.
The flagship quad core mobility Broadwell-H parts are still yet to be seen. There's wide-spread industry speculation around whether these traditional notebook parts will ever come to market. Industry rumors suggest that Intel may cancel Broadwell-H due to Skylake coming so soon afterwards. Skylake is scheduled to land in the back to school time frame with both mobile and desktop variants.
Intel's desktop roadmap still lists both Broadwell-K and Skylake-S parts to be launched simultaneously. We're unsure if this is still the plan as it seems quite redundant to launch two microarchitectures simultaneously especially when one is superior to the other in every aspect. The only two reasons that i could think of is if Intel plans to appeal to 9 series chipset owners that are looking for better integrated graphics performance, a small niche in my estimation. The other is the need to find a market for the leaky Broadwell parts that can't be used in their respective mobile platforms.
The more interesting bit in the Intel conference call is about Cannonlake, Intel's first 10nm part. While Intel never mentioned the part by name in the earnings call they did talk about the 10nm process. Krzanich indicated that the company still hasn't established the timing for the process. This makes a 2016 debut very unlikely. If the new process proves to be as problematic as 14nm then we're quite possibly looking at another Haswell like situation with Intel refreshing Skylake parts in 2016. It's too early to tell for sure but it's evident that the whole semiconductor manufacturing industry is slowing down, not just Intel.
While Intel's 14nm troubles did allow its competitors to catch up, especially Samsung who's been successful in ramping 14nm production recently. It's not the only company that's been facing issues with manufacturing technology. TSMC has repeatedly faced major delays of its own. Although it's always fun to play with a product featuring the latest and greatest manufacturing tech. We've come to learn that better products and innovation can still be brought to the market without the latest and greatest manufacturing technology. Nvidia's Maxwell and AMD's Carrizo are great examples of this. AMD's founder, Jerry Sanders, had a great analogy for this. He used to say it's not the photochemical film that's valuable, it's the movie that you shoot on it. As such manufacturing technology is just a means to create something of value.