Intel's 14 Nanometer: It's Here And It Kicks Butt

Sep.15.13 | About: Intel Corporation (INTC)

At the very first keynote by new Intel (NASDAQ:INTC) CEO Brian Krzanich, I had a front row seat to the demonstration of the world's very first, fully-working 14 nanometer microprocessor. Folks, this isn't some "test chip", but a bonafide, fully-functional, Windows-booting microprocessor that is set to go into production by Q4 2013. Thanks to Anand Shimpi of Anandtech, we have a close-up shot of this beautiful 14 nanometer silicon:

(click to enlarge)Click to enlarge

On the left (circled in red), we have samples of Broadwell-ULT (for Ultrabooks) and Broadwell-ULX (for fanless tablets). Now, to be crystal clear, the longer piece of silicon on all of these chips is 14nm (and 22nm in the one on the far right), while the smaller one at the top of the package is built on Intel's 32nm process.

Anyway, so there are a number of observations to be made here that are worthwhile from the perspective of the shareholders.

Honey, I Shrunk The Die

While the foundries are talking about how they do not expect much in the way of die shrinks for their 16nm/14XM processes and that the major advancement will be the move towards FinFETs, Intel's 14 nanometer (second generation FinFET) is a full-shrink from 22 nanometer. If you recall, I proved here that Intel's 22nm was a full shrink from the 32nm node (and Intel's 32nm node is indeed comparable to the rest of the industry's 32nm).

Note the CPU on the far right - that's Haswell-ULT built on Intel's 22nm process. Now look at the chips on the left. Not only does the 14 nanometer part have more transistors (the core is beefed up a bit and the GPU is likely to be substantially improved), but it's also about 64% the size of the Haswell. This, folks, is yet another "full shrink", which validates Intel's claims that it would again drive down the cost per transistor in its next process generation.

Bottom line is that next year, "Broadwell" on 14 nanometer will not only be lower power/higher performance, but it will feature a significantly smaller die size, meaning that there's a real economic benefit (i.e. gross margin improvement) to be had here.

Lower Power & Higher Performance

Smaller die is nice, but you know what's even better? Lower power. At the keynote, Intel showed a Haswell-ULT system and a Broadwell-ULT system running the exact same benchmark (and claiming that the test was "performance normalized" meaning that the chips were set to perform the same). The power numbers were incredible - a 30% improvement in power consumption at the "same performance". Folks, this is an Ultrabook chip that when running an intense floating point CPU workload, consumes under 5W for the entire system-on-chip.

Now, you'll notice that I put "same performance" in quotes - these guys were sandbagging hard. As I watched the live keynote, I noticed that they started the Haswell system first but by the end of the demonstration the Broadwell had actually made more progress on the particular benchmark!

(click to enlarge)Click to enlarge

Conclusion

The 14 nanometer Broadwell chip will go into fanless Ultrabooks and fanless high end tablets, which will be a pretty amazing feat for a high end processor with similarly high end integrated graphics. This is what 14 nanometer enables, and this is why the process technology advancements are key. Of course the chip designers deserve a round of applause (my contacts tell me that the design team actually found a number of neat ways to improve density even beyond what the 14 nanometer shrink brought), but at the end of the day, without process advancements, chip vendors run out of ways to improve performance and power really quickly.

While "Bay Trail" was really the star of the show at IDF for tablets, it's not going to be too long before the high end Core products make such good tablet processors that the Atom will largely be relegated to budget 10" tablets and to 7" devices, while the Core will own devices at the 10" and above form factors. At the end of the day, Intel has the entire compute continuum covered, and I can't wait to see what other goodies the company has in store for us over the next year.

Until then, bring on the Bay Trail sales!

Disclosure: I am long INTC. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.