Taiwanese Manufacturers Complaining about Intel's Prices. Again

Fresh news out of Taiwan is that Intel and AMD, despite their keen interest in the tablet market, "cannot compete with solutions from ARM in terms of price." The statements, purportedly from various notebook manufacturers, are meant to imply that the two manufacturers (especially Intel) must reduce costs if they ever hope to gain any traction in the market.

Then the article at DigiTimes coughs up this gem:  "Intel debuted 32nm-based Medfield processors with power consumption reduced to 11W to match Android 4.0 in January 2012 and will launch Atom processors with power consumption of less than 10W specifically for use in smartphones and tablet PCs in the second half of 2012, the sources indicated."

DigiTimes does some good work, but literally everything about that statement is wrong. We were briefed on Medfield in person at Intel HQ and spent time with prototype devices that were similar to the final products that'll ship out this year. The idea of a 10W phone -- any phone -- is ludicrous. The Saltwell core at the heart of Medfield uses 750mW of power at 1.6GHz; Intel's power consumption figures indicate that the entire SoC uses ~800mW when browsing the web or making a phone call. Even if we assume Intel's figures are better than what we'll see in shipping devices, there's no accounting for a 9W discrepancy.

Furthermore, Medfield is designed "specifically for use in smartphones." That was sort of the point. Intel is planning to follow up on Medfield with a new, Windows 8 tablet platform, dubbed Clover Trail, but the idea that the company needs to wait for 22nm to push below 10W is absolute FUD.

As for pricing considerations, we've discussed this topic as it relates to ultrabooks before; attempting to reframe the same topic in terms of tablets is disingenuous. AMD may have made a lot of brave noises about "Brazos 2.0" at its Financial Analyst Day last week, but it doesn't change the fact that the company's 4.5W "Hondo" APU is far too power-hungry to compete with the current crop of ARM tablets. Hondo is important because it's a stepping stone towards the company's SoC aspirations, but it's not going to compete effectively against the bumper crop of 28nm ARM chips and Intel's 32nm Atom.

x86 tablets are, in general, a non-starter in 2012. By the time Windows 8 ships very late this year / early next, AMD should be getting closer to having 28nm chips and needing to worry about trying to price them to compete with ARM. Manufacturers are planning to position Windows 8 tablets at a higher price/performance tier as compared to their ARM counterparts, much the same way that netbooks once divided between Linux and Windows XP.

AMD will undercut Intel as soon as it has parts that can fit into the same power envelopes, which is why Sunnyvale's "ultrathin" initiative aims to deliver ultrabook-style form factors at significantly lower prices. While AMD has little choice but to compete on price, neither it nor Intel are likely to start racing for the floor where margins are concerned. Both manufacturers have good reason to jockey for position as providers of premium performance product, at least until it becomes clear that price is the distinguishing factor that leaves customers turning to ARM instead of x86. That's not a discussion anyone is equipped to have until at least Windows 8.