Bricks and Mortar acquisitions are simple to conceptualize. When Ford buys a smaller car company, we assume that the other company will carry on producing basically the same products they’ve been making for years, only under the new banner. In many cases, the product line is not even re-branded, and many customers will not even be aware of the change. Administrative responsibilities are re-shuffled, corporate accounting reports look different, and, beyond that, everything appears to be business as usual from the outside.
Technology acquisitions, however, particularly those involving products like software or hardware IP, can give us that ice cream headache if we think about them too hard. Many times, the motivation for the acquisition is what throws us off the track. Some acquisitions are made simply to eliminate a competitor from the market. In other cases, the active customers are the valued commodity. Other times, the talent of specific groups of employees may be the golden goose. Or, the object may be a patent portfolio that conflicts with a strategic technology direction. It is rare to see a software or hardware IP company acquire another for the old-fashioned reason – just to make more money.
However, MIPS’s recent acquisition of analog IP supplier Chipidea seems to be just that, and the customers will probably come out ahead as a bonus. MIPS has long been a leading supplier of IP for system-on-chip (SoC) designers. According to the company, prior to this acquisition, MIPS was the third largest design IP supplier, behind ARM and Synopsys. The acquisition of Chipidea, the company says, puts them now at number two. Chipidea had claimed the number eight in that overall design IP market and the number one slot for analog IP. Why would we design engineers care about all this ranking and rating? Are we having an office pool for something more technical than sports? Probably it just gives us something to pressure the salesmen with when they come to call. “Well, I realize that you’re number four, but the number three company was just here yesterday, and they said…”
On a more relevant note, the MIPS/Chipidea acquisition is a good one for customers because there is no product overlap between the two companies. Why is that good? Because acquisitions with product overlap are bad for customers. You get that letter that says something like, “Dear Customer, We are happy to announce that, as a result of our acquisition, which included grossly overlapping product lines, we’ll soon be upgrading your [product that you love and use everyday] to our latest version of [product you’ve never wanted and will end up hating]. Please bear with us as we work to make this transition as smooth as possible. By the way, don’t count on your old applications engineer from [product you love company] to help achieve that smoothness, because he was just laid off to reduce costs. Have a nice day.”
These guys will not be sending you a letter like that.
With digital devices delivering billions of transistors of capability at sub-90nm geometries, integration no longer means just sticking more of your digital IP onto your ASIC SoC. Chances are, you’ve long since finished that project. The only digital stuff hanging off the edge of your big chip are things that are going to change often (like commodity memory, bridge logic you’ll stick into a programmable device, and maybe PHY devices for standards that weren’t stable when you did your ASIC). You’ve probably long since swung toward the “buy” end of the make versus buy decision for IP like processor cores and standardized peripherals.
MIPS has made a big business capturing that part of the market – particularly in the home entertainment electronics segment. The company claims huge market shares for processor IP in applications areas like VOIP (72%), DTV (60%), cable set-top-boxes (76%), IPTV (70%) and WiFi (55%). If you walk into a best buy or Fry’s, chances are you’ll find a lot more MIPS processors in the building than those made by, say, Intel. A huge portion of ARM’s number one market share in processor IP is from its dominant share of the mobile phone market. MIPS owns a large number of the other application areas.
Now that these companies put processors in most every product you design (and buy), they need a new way to make money helping you design these things faster. Already, both ARM and MIPS have expanded from offering pure processor IP to promoting platforms with broad ecosystems, including interconnect and bus structures, peripherals, memory, and just about everything else digital you can put into a system-on-chip to create a robust computing platform. Analog is the next logical step. With this acquisition, MIPS’s vision is to serve the same SoC design customers with both processor and analog IP.
MIPS’s push into analog IP may be well timed. As we move to smaller geometries with our digital circuitry, analog doesn’t scale the same way. As a result, an increasing proportion of the chip area, power, and design time are being consumed with analog – even for the same relative functionality. Usually, however, that analog portion of the design isn’t what makes it special. Our differentiation in most chips comes from the digital side. Combine these factors with the scarcity of qualified analog engineers and you have a perfect argument for outsourcing as much of the analog design work as possible. MIPS claims that analog is the highest growth opportunity in the SoC IP market today. Of course, they just voted with millions of dollars so we kinda’ already knew they’d say that.
When the analog portion of our designs was small, not very complex, and not on the critical path, it made sense for many of us to design it ourselves. We took a couple of analog classes in school, right? That RLC stuff can’t be all that complicated. Unfortunately, some areas like power management, high-speed analog to digital conversion, the PHY portion of high-speed serial interfaces, and on-chip RF are not the purview of the digital engineering tourist. Issues like signal integrity and power minimization require those guys that, you know – do this for a living. Chipidea got a couple hundred of them together a few years ago, which, by our calculations, is a good percentage of the world’s population of those people. (OK, how many analog engineers does it take to change a light bulb? – Oh, never mind.) Those analog gurus have been working together to crank out very robust implementations of the most popular and challenging IP blocks that most of us need in our SoCs. Power management, high-speed converters, high-speed interface PHY blocks, and RF are the staples of Chipidea’s 400+ block catalog.
One of the most time-consuming (and least rewarding) aspects of developing your own SoC analog IP is porting it ahead when you move up a process generation. Because it doesn’t scale as easily as digital IP, analog can become the long pole in the schedule tent, even if it wasn’t on previous turns of the chip. Many companies now opt to buy IP for blocks they previously designed, simply from a risk-reduction perspective. This played right into Chipidea’s master plan, of course, at least until it became a part of MIPS’s master plan.
For those of you that pay attention to the financial side of things – get an accountant already. You’re engineers, for goodness sake. If you must know, however, the acquisition was cash (meaning, I think, that your MIPS stock didn’t just get diluted). The Chipidea income will now just add to the MIPS reports. How clean is that? It’s just like when you buy a new piece of IP and wire it up to the rest of your design, and…