Monday, 9 May 2011

I Told You So! (Episode 2)

Way back in November, in a blog article called "ARM's Cortex-A15 CPU, and how it will change your world"", I closed with this paragraph:

So here's the real left-field question: given the power and BoM cost savings, how long now before Apple ditches Intel completely for ARM?

In a follow-up ("Desktop ARMs - and what they'll mean for Microsoft"), I expanded on those thoughts, opening with:

If Apple finds that the ARM processors out-compete Intel on price and power consumption, and at least match x86 for performance, the choice to switch is going to be a no-brainer.

And lo, it came to pass. Again.

Today, Charlie Demerjian published a very interesting article entitled "Apple dumps Intel from laptop lines" in his own blog, "Semi Accurate".

It seems as though I've been a tad prescient. That, or well-informed.

The rest of my blog article contained some other, equally dramatic, predictions. Well, let's call them "predictions" for now. One, which I've since admitted came from insider sources (in the first "I Told You So!"), was that Microsoft was porting Windows and Office to ARM. There were several more that are waiting to be proven.

Now here's another, just to keep your interest whetted. Rik Myslevski has written a well-informed article for The Register, titled "Intel's Tri-Gate gamble: It's now or never". In it, he adds a throw-away speculation near the bottom of the last page: "Intel could license the ARM architecture and start buiding its own ARM variants in its own fabs, using its 22nm Tri-Gate process. That's unlikely, but stranger things have happened."

Stranger things indeed. And I think that Rik is well and truly on the money. Intel's executives are very, very far from dumb. They are painfully aware of ARM's squeeze on their markets, both from above and below, as I'd outlined in that first blog article. Intel really has three obvious plays they could try:

1. Find a transformational technology that puts them in contention with ARM but retains x86 compatibility.. Intel's Tri-Gate announcement was clearly an attempt at this play.

2. Try to find a way to put ARM out of business, or at least weaken it. We've not seen this attempted seriously yet, and it's reasonable to wonder why that is.

3. License ARM cores for desktop use, and keep in the market by following the market..

I contemplated in the second blog post:

Here's a statement I never thought I'd make: with a stock and cash exchange, Apple. Could. Buy. Intel.

Now things start to add up.

* Apple needs a top-tier ARM supplier. Their relationship with Samsung, their current ARM fabrication partner, is reportedly getting a little rocky. Industry reports suggest that they're considering switching to Intel chip fabs.

* Intel needs an answer to the ARM squeeze on x86.

* Apple has tons of ready cash.

What it all adds up to is the possibility that Apple could invest in Intel, both commercially and financially, to license the ARM cores, and set up new fabs to make Tri-Gate ARMs, for both their iOS and MacOS products.

It would make a great deal of sense. It would also give Apple the power to put the squeeze on Microsoft. If only Apple has access to the Tri-Gate ARMs, it leaves Microsoft out in the cold - or rather, the far-too-warm - when MS start to produce the ARM versions of Windows, which will only be able to run on old-school pre-Tri-Gate hardware.

Microsoft will have to consider a very similar investment in Intel, for the same reasons, before Apple can lock them out.

Whichever way you look at it, Intel's prospects may well have brightened.

LATER NOTE (2011/05/18): it seems that one or two people may have read this blog, after all. :) Paul Otellini of Intel was put on the spot in an investors' meeting about the possibility of Intel using Tri-Gate to fab ARMs, and did his level best to dampen speculation: "The short answer is 'No'."

Mind you, if Apple came knocking, with barrow-loads of dollars in tow...? It's possible that Otellini's playing a wooing game, as his next comments could be construed as a come-on: "We have [...] an ARM architecture license. The important thing for us is to figure out how to get paid and how to be present. And we think the best way to be paid and present [...] is to build best-of-class chips."

EVEN LATER NOTE (2011/05/28): perhaps Paul Otellini wasn't reading from the hymn-sheet after all...Intel's CFO Stacy Smith appears to be a lot more positive to the idea of Tri-Gate on ARM, and specifically mentioned Apple as a possible buyer (of devices, not Intel!), according to a Reuters story this week.

Now, bear in mind that a CFO's words probably carry even more weight in the financial markets than the CTO's. After all, CTOs have been known to make some pretty daft comments - but the CFO is the person the money listens to. Smith said that an Apple deal was "Not in the works today". To the untrained observer, that's a flat denial, but it's more likely to be coded speech for, "We're not manufacturing yet, but a deal's cooking." We shall see. Keep watching this space!

No comments:

Post a Comment