Intel paying $884m for Wind River?
IBM should make a counter offer
Comment Financially struggling chip maker Intel is paying $884m to take over embedded operating system supplier Wind River Systems. Why would it do that, you may ask.
Well, the best answer, of course, is because it can. Intel's launching and continuing support of the Moblin variant of Linux for netbooks shows that the chip maker knows that the choice of the operating system and application stack that customers make often determines the choice for the underlying chip and chipset that gets used.
Intel may not want to be in the operating system racket, but it didn't want to be in the compiler business, either. And it has had little choice but to put together a decent stack of compilers to help software developers - as well as OEMs and ODMs - take advantage of all the new features in each successive generation of Intel processors.
Intel's Atom processors are in the fight of their life for consumer embedded projects. Stacked against them are a slew of creative licensees of the ARM RISC processor and Advanced Micro Devices putting out decent low-voltage parts that makers of embedded systems are fond of. Intel is going to need a concerted hardware-software attack to preserve, much less extend, its presence in the embedded systems arena.
The fact that Intel was willing to spend a 44 per cent premium over yesterday's closing price for Wind River shows just how serious Intel is about expanding out beyond its core PC and server businesses. It is good for a public company to be serious, but Intel also had aspirations in systems management and networking, to name two areas that didn't pan out. And the embedded systems racket, as Wind River knows full well, is tough. The margins ain't so great, either.
But, with IBM yammering on about smart infrastructure at every opportunity these days, and the big governments of the world beginning to invest in such projects, the margins for embedded operating systems could soon improve significantly. The irony, of course, is that the embedded variants of the PowerPC processors - from IBM and Freescale - are popular for embedded systems. Intel now controls one of the most popular software environments for those embedded systems.
Maybe IBM's Rational division within its Software Group will make a counter offer? Pity its merger and acquisitions expert is trying to move to Dell.
The fact that Wind River, which was founded in 1981, has deep experience in embedded systems perhaps makes it worth the premium that Intel is paying, and worth IBM starting a bidding war for it. Wind River sells two different platforms, a carrier-grade VxWorks Unix-alike operating system as well as a homegrown carrier-grade Linux distro originally called Carrier Grade Linux and now called Wind River Linux 3.0. (Sun Microsystems is now using Wind River's embedded Linux on its Sparc T series of processors after commercial Linux distributor Canonical pulled the plug on the Sparc T chips back in October 2008.)
Both VxWorks and Wind River Linux are made useful through the Workbench application development environment that comes with the platform. But Wind River also has something else of value to Intel: experience helping companies use the Android handset variant of Linux championed by Google.
How Intel will mashup its compilers and Wind River's tools remains to be seen, of course, and Intel won't say anything until the deal closes this summer.
Intel said that Wind River will be a wholly owned subsidiary of Intel and after the deal does close, Wind River will be tucked up into Intel's Software and Services Group, headed by Renee James. The embedded operating systems and development environment will be a crowbar that Intel uses to try to break into handsets, infotainment, networking gear, process controllers, and other places where embedded applications run, and which Intel characterizes as a "multi-billion market opportunity."
It had better be, at the price that Intel just paid for Wind River. It seems fair to assume that Intel will be cutting some of Wind River's 1,600 employees from the ranks in the back office to make it a more profitable part of the chip maker, once absorbed.
In its fiscal 2008 year ended in January, Wind River had $359.7m in revenues, but only $10.7m in net earnings. In 2007, Wind River had sales of $328.6m, but booked a loss of $2.4m, and 2006 was not really much better, with sales of $285.3m but a net income of only $573,000.
Wind River has shown decent top-line growth, but has not been able to bring it to the bottom line at the rate that a software vendor should be able to - especially one that is part of the military industrial complex. ®
> "Wind River is buying out your RTOS vendor"
Yes, I remember that too. Still, at least they weren't saying that you had to change processor too (at least not very much).
Just another iteration
I wonder if Intel will do to Wind River what Wind River has done with everyone they have bought out. About the worst news an embedded developer can get is "Wind River is buying out your RTOS vendor" since you get to start all over again.
I think you mean "QNX's Neutrino" and "IOS-XR", so er... no. VxWorks has plenty of successful implementations, but that isn't one of them.
IOS XR does use QNX, never heard about KR thing though (except the K&R C of course).
They don't know what they're getting
There's an awful lot of stuff out there with an awful lot of investment behind it that uses VxWorks on chips that aren't from Intel. They're using ARM and PowerPC, even 68000! And those things and their developers with decades of ecosystem behind them aren't about to switch to an Intel chip just because VxWorks has been bought, even if Intel withdraw VxWorks-PowerPC, VxWorks-ARM, etc.
If Intel do withdraw that support I expect a lot of things will migrate to Greenhill's Integrity OS. It's not too dissimilar to VxWorks, and does support all those other CPUs that don't belong to Intel.
Is this all a symptom of Intel having ditched XScale, regretting it, not admitting the mistake, trying to cover it up with their creation of Atom, realising that Atom is a mistake too, and trying to cover that up by buying a key OS vendor with a view to force developers on to Atom? They'll have to buy all the other real time OSes too, and that'll look anti-competitive...
Intel have a lot of work to do to make x86 fit the same niches as ARM and PowerPC. Atom doesn't even come close.