SAN JOSE, Calif. – I've got a theory why Intel is struggling to get a foothold in TVs and cellphones—it's a lack of humility or, put less kindly, hubris.
While I was taking a few days off last week, Intel quietly announced it was getting out of the digital TV business. It had produced two generations of SoCs--one for Yahoo software and another for the high-profile GoogleTV initiative.
In 2004, the PC giant abandoned a separate digital TV effort that tried to leverage microdisplays, an emerging technology that never quite emerged.
Only with GoogleTV did Intel have solid OEM partners—in this case Sony and Logitech. In this case broadcasters put up barriers to the notion of Google eating what's left of their advertising lunch.
Web-enabled TVs are coming in various forms. But it doesn't look like they will be using x86 cores anytime soon.
The problem, in part, is Intel seems to take a bulldozer approach to new market development. I give the company credit for its willingness to drive ambitious initiatives aggressively, but too often it does that without key stakeholders such as OEMs, or in the GoogleTV case, broadcasters.
The same has been true in smartphones. Years ago Intel developed one of its first cellphone SoCs, merging x86 cores and flash on a chip, but it never got market traction. Instead 3-D chip stacks with separate flash and application processors took off from traditional suppliers.
In a second effort, Intel has been driving Atom into the smartphone market for about two years. It signed up Nokia as a development partner, but incoming Nokia CEO Stephen Elop nixed the deal when he shifted support to Microsoft, perceived as a more vibrant ecosystem provider.
Intel is still beating the drum for its 32nm Medfield SoC based on Atom, claiming next year it will have multiple smartphone design wins. But it also cancelled its MeeGo flavor of mobile Linux originally geared for those phones.
One source tells me Intel warned some of Taiwan's big ODMs as recently as three months ago they were not putting enough effort into MeeGo. A handful of small software companies were working with the OS and now feel they have wasted more than a year's time. Intel is encouraging them to shift their focus to Tizen, a replacement mobile Linux OS coming next year.
I give Intel credit for quickly winding up big, bold initiatives, and having the flexibility and guts to trash them when they don't pan out. For a big company, Intel moves quickly.
But the knock-on effect is potential partners get shell shocked, unwilling to jump on the next big bandwagon. Among other past misfires: wireless USB, Advanced Switching Interconnect and Intel's network processors sold off to Netronome.
Yes, there were also roaring successes such as PCI and PCI Express. And Intel has made great strides getting the x86 into diverse embedded systems, including control plans in comms systems. But I cut OEMs and other partners some slack for being reticent about the next big Intel initiative.
I suspect Thunderbolt may be the next to go. So far, Apple is the only top OEM supporting the interconnect which overlaps USB 3.0, a link most PC makers have been working to support for years. Intel claims Asustek and Acer will adopt Thunderbolt next year, but I suspect the support may not be very broad across their product lines.
Indeed, Taiwan has been so complaint with every major Intel initiative it is almost complicit in fueling the x86 giant's hubris that it can boss its way to success. These days even top tier PC makers are so gutted of technical resources they rely more heavily than ever on Intel, adding swagger to the microprocessor makers stride.
Intel badly needs to get key sockets in tablets and smartphones. Bulldozing Atom with whatever will be the new mobile Linux flavor of the day may look like the shortest path, but it is not the best one.
I think Intel should let its new wireless baseband group acquired from Infineon lead the way. The group should be a strategic listening post with ears deep into the operations of key handset makers, including Apple. Listen to what those OEMs want, what their struggles are. Look for a gap and when you find it, fill it quickly and cleanly.
That's an approach that requires a lot of humility, letting customers lead the way. Most of the semiconductor industry has been using that model for years.
If Intel wants to get beyond its browbeaten customers in the PC market, it needs to learn these widely practiced techniques and develop the patience to take the long road.