Imagination announced that it will restructure the company to cut costs, which will lead to another 200 layoffs in addition to the 150 announced last month. The 350 employees would reduce the company’s workforce by about 20 percent of the current 1,700-strong workforce.
Out With The Bad
The company’s operational costs will drop by £15 million ($22 million) for the financial year ending April 2016. Imagination will also cut non-core or cash-consuming units, which will bring an additional £12.5 million ($18 million) in savings delivered in full in the financial year ending April 2017.
Some of the business units being divested include the consumer radio business, the System-on-a-Chip (SoC) design business, as well as other overheads that weren’t focused on the core businesses of the company.
Refocus On PowerVR, MIPS, Ensigma
The company will not cut any jobs from the PowerVR, MIPS and Ensigma (connectivity) businesses. In fact, it’s looking to hire 50 more employees for the PowerVR unit. The PowerVR business is the most profitable unit, so it makes sense to strengthen it.
Ensigma has also been a growing business for Imagination, and a unit in which the company has kept investing to improve it. Imagination bought MIPS not too long ago, so for now, the company believes it wouldn’t make much sense to get away from it. That could be a mistake, which we’ll address later.
When a company starts cutting jobs, it’s usually not necessarily because the company has spread out too thin (although that can be one of the factors), but because the main business isn’t as profitable as it once was. That means the same projects that the company started and could sustain in the past have become unsustainable.
Although Imagination has done very well for itself in the past decade by selling GPU designs to a fast-growing company such as Apple, it also made it highly dependent on Apple’s success. If Apple’s iPhone business goes down, then so does Imagination’s GPU business.
This is made worse by the fact that Imagination has been steadily losing GPU market share in Asia, where chip makers have begun to adopt ARM’s Mali GPU or GPU designs from other competitors. Qualcomm has also eaten into Imagination’s GPU business simply by selling more SoCs (which include the Adreno GPU) to Asian smartphone manufacturers.
Although it’s not clear why Asian chip-makers would prefer Mali over PowerVR, as PowerVR is still one of the best mobile GPU architectures on the market, it likely has something to do with pricing.
Imagination may not be willing to cut its price for Asian customers, because that would probably mean having to cut prices for Apple as well. However, if that's the case, the company may have pushed chip makers into ARM’s, er, arms.
Imagination bought MIPS Technologies for $60 million at the end of 2012 -- a sum that’s not exactly small for an IP company such as Imagination. Also, at the time, MIPS was losing money on an annual revenue of $60 million.
Imagination wanted the MIPS CPU architecture possibly because it thought it could promote a third major instruction set architecture (ISA) that would be on a more equal footing to ARM and x86. It may have also worried that ARM would eventually close down its ISA to only work well with its own GPUs.
More than three years have passed, but so far we haven’t seen any major growth for the MIPS architecture. For the most part, this is probably because there’s no room for another proprietary ISA, but it also feels like Imagination hasn’t been aggressive enough in promoting it.
For starters, Imagination hasn’t shown that it can compete in peak performance against ARM. It has tried instead to compete mostly at the low-end and mid-range and in “performance/Watt” metrics. That’s a worthy goal, but it’s not going to get the company the headlines it needs to make MIPS successful.
AMD, for instance, lost the battle with Intel mainly because it couldn’t keep up with the peak performance of Intel’s high-end chips. Those chips didn’t represent “just another pricing level” - they were a representation of Intel’s chip leadership and innovation.
Later, everyone started believing AMD’s chips were generally “worse” than Intel’s chips, at all pricing levels. In the end, that actually became true, because AMD had fewer and fewer resources to put into its chip business.
Imagination has also failed to gain a chip-making customer that would create a high-end MIPS CPU for smartphones on cutting-edge process nodes. Getting such a customer likely wouldn’t have been easy, as most smartphone OEMs are invested in ARM, and probably not too many were willing to take a chance on MIPS. However, this just shows what an enormous uphill battle Imagination has had with MIPS -- a battle that may not get any easier going forward.
RISC-V: A Better Bet Against ARM
If it would’ve waited two more years, Imagination could’ve gotten its own ISA for free. RISC-V is an open source instruction set architecture, with version 2.0 being publicly released in 2010, after being introduced by UC Berkeley in 2010. Imagination could’ve created its own CPUs on top of an ISA that’s not controlled only by ARM.
The difference between MIPS and RISC-V is that due to its open source nature, it could actually gain much more adoption than MIPS in the future from chip makers, much like Android got high adoption from smartphone OEMs. Even though it’s very early, some big names such as Google, HP, and Oracle are already backing RISC-V, and it’s possible more will adopt it in the future.
Even Intel couldn’t make big gains in the mobile or embedded market against ARM, because no matter how rich it is, ARM’s business model is empowered by many companies adopting its architecture and making chips on top of it.
MIPS could’ve worked the same way, but it’s been a few years since Imagination got it, and it doesn’t seem to have worked. RISC-V’s chances against ARM aren’t guaranteed, either. However, if there’s any other ISA that can take on ARM in the mobile and embedded markets over the next decade, that’s probably RISC-V.
Near-Term Goals For Imagination
The thing Imagination needs to do right away is try to win back Asian chip makers and pull GPU business away from ARM, whether it’s by offering low-end competitive GPU IP that costs significantly less, or through other means.
Because Qualcomm still dominates the vast majority of high-end smartphones, that means Imagination will need to win companies such as MediaTek or Huawei and get them to use its GPUs.
Then, they can work together with smartphone OEMs to put those chips into popular devices and promote those devices as better at gaming, or more efficient with high-resolution displays, or what have you. Imagination clearly needs some strong brand promotion here as well. Customers also need to want PowerVR GPUs in their phones, over anything else.
If Apple’s iPhone business will indeed start to slow down, as some analysts are predicting, then these goals become all the more urgent for Imagination. The company would not be able depend on Apple to increase its growth anymore.
Lucian Armasu is a Contributing Writer for Tom's Hardware. You can follow him at @lucian_armasu.