For quite a while I have been critical of chip giant Intel's (INTC 3.28%) execution vis-a-vis chip manufacturing technology. The company has long touted its leadership position relative to its major competitors, but over the last few years, Intel's execution here appears to have faltered as said competitors seem to have substantially improved their own.
Intel's execution issues in chip manufacturing can essentially be boiled down to two things:
- Intel hasn't been able to launch fundamentally new manufacturing technologies at the rate that it had done previously, while competitors seem to be moving more quickly.
- When Intel does finally bring a technology into high volume, it seems to start production at lower manufacturing yields (aka higher costs) than it did in the past, indicating even further "delays" normalized for yields.
After Intel hosted its Technology and Manufacturing Group (TMG) Day back in March, and after some new information leaked out about the company's upcoming product schedules, it's worth checking in to see if Intel has made progress in shoring up its execution in manufacturing technology.
14nm+ and 14nm++ seem to have gone well enough
Intel has made it clear that it aims to deliver multiple "waves" on a given chip-manufacturing technology. The idea is that the company expects to take longer to deliver new manufacturing technologies with substantially reduced chip area compared to previous ones, but in the interim, Intel's product teams can still deliver better products through a combination of design enhancements and manufacturing technology performance boosts.
In the second half of last year, Intel launched its first seventh-generation Core processors (often referred to by its codename, Kaby Lake) using its 14nm+ technology.
The technology was quite good (Intel claims a 12% performance boost over its first-gen 14-nanometer technology), enabling sizable performance enhancements across many of Intel's product lines (particularly those that are power constrained).
The company's first products based on its 14nm++ technology -- which Intel says will offer another double-digit jump in performance -- should begin to ship in the third quarter of this year.
Indeed, both Intel's upcoming Kaby Lake refresh chips for notebooks, as well as its first Coffee Lake chips (aimed at the enthusiast gaming market), are expected to launch in the third quarter of this year. That's about a year after the first 14nm+ products made it to market.
The jury's still out on 10nm, though
Intel has done a good job of consistently executing on derivative technologies based on its original 14-nanometer manufacturing technology, but before I'm ready to confidently state that Intel's execution has improved, I want to see how it does with its 10-nanometer technology.
The transition from any 14-nanometer derivative to 10 nanometer should be much more difficult (and much more indicative of management execution) than the transition from any one 14-nanometer derivative to another one.
If Intel can smoothly deliver products based on its 10-nanometer technology and then follow them up relatively quickly with 10nm+ and 10nm++ based products, then that'd be solid evidence that things have really begun to turn around for the company's manufacturing division.
However, if the 10-nanometer product rollout isn't smooth and is plagued with issues, then that'd be a clear sign that the company's manufacturing group is being poorly managed and ripe for a shakeup.