Intel has a big financial problem with Coffee Lake and it isn’t the mediocre specs. As SemiAccurate has been chronicling for months now, it is about their cost structure in light of the continuing process failures.
We told you about 10nm yields with our exclusive story about Cannon Lake’s tapeout and first silicon. We then explained why 10nm was not just broken in terms of yields but in terms of quality of product, either of which can cripple the CPUs produced on it. The fact that Intel was struggling with both wasn’t a good sign.
Then there was the Intel Manufacturing Day last April. In it they introduced a new term, “Hyperscaling”, which is supposed to convey that although shrinks had lengthened from two years to three, Moore’s Law was still alive and well. This is the long way of them claiming that they aren’t actually screwing up with a 2+ year delay on 10nm. If you look at the numbers presented, the numbers not presented, and the fact that financial projections were based on the rather curious data points, it all becomes clear as to why Intel is so desperate to distract onlookers.
Coming back to Coffee Lake, SemiAccurate gave out the specs we found earlier today. The speeds and feeds are about what you would expect from a warmed over Kaby Lake, same process, roughly expected speeds, and launched without enough facts. Coffee appears to underwhelm on performance but the real problem is what it means for margins. Intel no longer gives out die sizes, transistor counts, or technical information for a reason, and Coffee Lake is another big reason.
Note: The following is analysis for professional level subscribers only.
Disclosures: Charlie Demerjian and Stone Arch Networking Services, Inc. have no consulting relationships, investment relationships, or hold any investment positions with any of the companies mentioned in this report.