Santa Clara (CA) – When Intel CEO Paul Otellini announced “everything with a bracket on it” would come under scrutiny in his company’s reorganization plans, the first official details of which were scheduled to be released next month, analysts began predicting how certain underperforming units could be broken cleanly away from the company. Many forecasted the outright sale of its Itanium processor unit; others suggested the sale of its NOR flash memory unit, which has a more difficult time these days competing with cheaper NAND flash.
So when the San Jose Mercury News broke the story that Intel was searching for a buyer for XScale technology, analysts had a difficult time finding the white dotted line that marked the point of separation. Just a few years earlier, even Gartner analysts were foreseeing Intel’s push in cellular phone technologies as leveraging its strength in processors to gain a foothold in a lucrative field. To some, Intel’s move into cellularlooked like a seamless transition, another sign that big and small technologies were converging.
It isn’t that Intel was necessarily unsuccessful in attracting and maintaining customers. BlackBerry devices, the Palm Treo, and the Motorola Q phone – the most powerful new designs in the industry – now depend on XScale processors. So some analysts and business magazine editors suggested that selling XScale would be wrong – that Intel should instead simply find a way to capitalize on its successes there. There didn’t seem to be an obvious seam, a “tear point,” a place where XScale could separate from Intel and still be viable for someone else.
One problem the Asian edition of Electronic Engineering Times foresaw dealt with intellectual property. Intel doesn’t really own XScale, it noted; it acquired a license to XScale from ARM Holdings, Ltd., as part of the deal years ago involving the distribution of assets from the former Digital Equipment Corp., which was acquired by Compaq (which was acquired by HP). XScale technology, some noted, involves the use of specific manufacturing processes which Intel developed, mainly for its own use, which some said could not be easily transferred to, or even compatible with, competitors’ technologies which are based on Taiwanese (TSMC) standards.
Some suggested that Intel should hold onto XScale because it still has value to the company with respect to WiFi, WiMAX, and next-generation digital networking. Selling XScale, they said, would create a tangle of intellectual property that prospective buyers wouldn’t want to bother with.
This Google Earth map of Santa Clara depicts the distance in feet between Marvell’s headquarters and Intel’s. Their positions on cellular technology aren’t much further apart.
It’s a strange wonder that all the analysts and editors didn’t pay close attention to the logic of their own suggestions. If they had, they might have seen the big, flashing beacons which pointed the way to Marvell Technologies. Despite what some of the headlines said this morning, Marvell didn’t purchase “the XScale unit” from Intel today. In other words, the “tear line” isn’t where people thought it would be or should be. In fact, the same intellectual property that analysts thought presented a tangle, actually pointed to a clear path to separation: Marvell today purchased Intel’s license to manufacture XScale technology, specifically for cellular and handheld devices, granted to Intel years ago by ARM Holdings. It’s a simple transfer of rights, with a bonus attached that Marvell gets Intel’s people who build it for them – up to 1,400 employees, whose paychecks will now be authorized by a building fewer than two miles away from the previous signer.
Meanwhile, Intel retains XScale for use in WiFi and WiMAX, which is exactly where analysts said it remained valuable to Intel anyway. Intel even retains the use of the XScale trademark. “It’s the cell phone area that we’re figuring is a better fit for Marvell,” Intel spokesperson Robert I. Manetta told TG Daily this morning,” and we should be focusing on technologies like WiFi and WiMAX.”
“We had design wins with the BlackBerry, the Palm Treo, the Motorola Q phone,” Manetta reminded us. “These products are XScale-based, that’s true, but this business [that Marvel is acquiring] is specifically the handheld business. There are other businesses within Intel that use XScale that are not part of this sale. We use XScale, for instance, in our storage business and also our network processors. Those are not affected by this sale.
“We made a decision that, even though we’ve had some success with this business,” Manetta continued, “it was not an area we wanted to focus on, and it would be a better fit at Marvell.”
Fit? With the alleged intellectual property quagmire ahead of any potential purchaser, how could XScale fit? A quick check of the Internet’s ever-accumulating archive of press releases would answer that question right away. In early March at the Intel Developer Forum, Marvell announced its products already supported XScale. No transition to TSMC was necessary.
There was also the momentum toward a Marvell purchase, which the world at large apparently missed. Marvell has spent billions on acquisitions in the last five years, beginning with gigabit Ethernet company Galileo Technologies in 2001 (which Marvell used to establish a customer relationship with Intel). In February 2003, Marvell acquired networking software provider RADLAN Computer Communications; then last December, Marvell acquired semiconductor design company UT Starcom. Finally, just last March, Marvell acquired the printer ASICs division of Avago Technologies.
So here’s a company that was clearly interested in expanding its interests in precisely the business that Intel signaled it wasn’t interested in anymore. And even its geographical distance from Intel is less than two miles. All that was missing to draw any more attention to this possibility were the dancing girls and a trapeze act. And yet we all missed it.
And some of us are still missing it. Intel stock traded down today, with financial analysts casting suspicion on Marvell’s ability to manage a money losing unit. Here’s the memo, people: They’re not acquiring it – not outright. Robert Manetta: “What they’re getting is: 1,400 employees, more or less; our current products and our future products that we’re currently designing into handhelds; and then our customer relationships that we’ve built up with names like Research in Motion, Palm, Motorola, and others.” What Intel gets is $600 million and, perhaps more importantly, a load off its shoulders, in the form of relief from some of its debt burden.
“[Intel’s] storage and network businesses,” Manetta also stated, “will be able to continue licensing that technology, and then alter it as they’ve always done for their specific applications. Marvell will be able to do that as well; since they’re acquiring this business, they’ll be licensing the technology from ARM, and then altering it in the way they see fit for their applications.”
It wasn’t supposed to be feasible. Yet it’s happening, unless some “I Told You So” tragedy happens during the next four to five months to unravel the deal. At this point, that doesn’t seem likely. If the remainder of Intel’s transition plan is as savvy as its deal with Marvell, some of us out here who are getting paid to observe may have a few more lessons to learn.