Intel resists Moore’s law. The company worked like a clock with that unstoppable Tick-Tock cycle that assured us lithographic jumps every two years, but those were other times.
We have been using Intel 14-nanometer processors for four years, and it will be at least another year until we see the first CPUs manufactured with a 10- nanometer process. Brian Krzanich explained the difficulties of a jump that is still resisting and that returns to make the image of this giant suffer.
Intel is doing well, or so it seems
The announcement of the delay occurred during the presentation of financial results for the first quarter of 2018. Intel may be satisfied, because its revenues continue to grow: they achieved 16,100 million of revenue (9% more than last year) and a net profit of 4,500 million dollars, no less than 50% more than the same period of the previous year.
The company does not do badly at all thanks to the diversification experienced in areas such as artificial intelligence, autonomous driving, 5G connectivity or promising projects like those dedicated GPUs that we all hope to see very soon.
In fact Intel is doing well in almost everything, at least according to those financial results. Its solutions for data centers, its business of non-volatile memories and even its division of Internet of Things grow at rates that are around 20% compared to the first quarter of 2017, but in these figures there is a big shadow : the one that plans on its division of processors for end users, which has grown by only 3% compared to last year.
The 10 nanos will wait again
Years ago Intel said that the 10 nanos would arrive in 2015. It happened 2015 and the company ended up indicating that these microprocessors would not arrive until 2017. It happened 2017, and once again they explained that they had more problems: do not worry, that 2018 will be the year of the 10 nanos. But it is that this will not be that year either.
Intel CEO Brian Krzanich said that they are already distributing some units in small volumes and production is improving, “but the rate of improvement is slower than we anticipated, as a result of which mass production goes beyond the second half of 2018 to 2019.”
Krzanich explained how the production was having problems that they had identified and apparently already know how to correct, “but it will take time to implement those changes”.
That will cause Intel to refine the manufacturing process of 14 nanometers to offer its future new family of microprocessors for both end users (Whiskey Lake) and servers (Cascade Lake).
It’s not good news, especially considering how the competition is. AMD competes more from you to you than ever with Intel in its microprocessors (we have seen it with its Ryzen 2), but it is also known that the company is already working on the development of GPUs with manufacturing processes of 7 nm in its laboratory, and will offer a preliminary version of the EPYC 2 with this process by the end of the year.
Nor does the fact that investment in R & D is relatively modest help. It has only grown 3% in this quarter in this area, and these problems – coupled with scandals such as Specter and Meltdown, although they have already been mitigated – again call into question the capacity of a manufacturer that seemed a safe bet. Only four years.