NEW YORK--Correct us if we’re wrong, but many readers may be writing off MIPS Technologies way too soon.
The fate of MIPS is shaping up as one of the industry’s biggest stories in 2012. The company’s potential sale would have a significant impact on its storied processor IP line, MIPS customers, potential suitors who would either gain access to the processor IP or lose access and, of course, the current MIPS work force.
MIPS has not commented on speculation about a possible sale, and we’ve been unable to confirm that it has retained an investment banker to represent it (the rumor mill says it’s Goldman Sachs). Nevertheless, the question of who might buy MIPS is unavoidable and should be debated considering the broader implications of such a deal.
For one thing, there are enough chip companies -- including Broadcom, Sigma Design, Cavium Networks and others -- whose SoC roadmaps are dependent on MIPS cores. The outlook for MIPS licensees is as muddled as the future of MIPS. Switching to new cores is tricky since it requires an extensive rework of the internal software infrastructure. As one EE Times reader noted: “Current-day SoCs need a fast turnaround time, and the time needed to rework the software / debug with a new processor IP is just not there.”
But some argue that the transition might not be all that traumatic. Gary Mobley, senior research analyst at The Benchmark Co., wrote in his recent research note that “there is evidence that many of MIPS’ long-time licensees such as Sigma Designs and MStar have started to migrate toward ARM cores.”
To close a deal, seller and buyer must first agree on the sale price. MIPS, which has reportedly been shopped around for the last decade, the question of price is a moving target. In assessing the value of MIPS, Mobley wrote that “the wild card is MIPS patents.” He believes MIPS’ patents, “at a minimum,” could be “worth $100 million.” Mobley estimates the liquidation value for MIPS at about $5 per share “based on our best estimate for the present value of the company’s royalty stream, plus cash and certain assets, minus balance sheet and off-balance sheet liabilities.
MIPS understands how to play this game.
J. Scott Gardner, a senior analyst at The Linley Group, said MIPS has had its patent portfolio evaluated and is now included in the Ocean Tomo 300 Patent Index for 2011-2012. Ocean Tomo, an IP merchant bank, developed an “IPQ score” to rate patent assets based on a proprietary statistical methodology. The IPQ score has a median of 100. Patent assets with higher IPQ scores are statistically more likely to generate economic returns, according to the firm. The IPQ score for MIPS is 135 – one of the highest among semiconductor companies listed in the Ocean Tomo 300 list of member companies
The list “includes a lot of very large companies, and the median market cap is over $9 billion,” said Gardner. “MIPS could raise cash without selling the entire company.”
Gardner added that there are “a few scenarios that would really shake up the industry,” including Apple and Google. “Apple has proven the value of complete vertical integration and ecosystem control,” he said.
Others find promise in the emergence of a tighter Google-MIPS relationship. Google has recently decided to include the MIPS architecture as part of its SDK for the next generation of Android, noted Mobley. "This means eventually all future Android applications will run native on MIPS.
In the following pages, EE Times editors weigh in on the debate, sharing their thoughts in the categories of: a few unlikely candidates, best bets and a potential wild card.
My bet in this speculative discussion is on Intel too. They desperately need an out into the mobile market; and with ARM infringing on the server market it makes sednse for Intel to use its cash to do with MIPS as they see fit. The question is, is there a tight fit between MIPS and Intel strategies? For a company that capitalized on a radically different RISC architecture developed by current Stanford U. president J. Hennessy MIPS might be just what Intel needs to counter the ARM-based mobile juggernaut.
With China's recent commitment to a national silicon standard (ISA), a Chinese firm (newly created, established, or government) makes perfect sense as a purchaser. I suspect that any regulatory and security concerns can be negotiated.
Good and interesting calls on Microchip and Intel, which we did not include in our runners and riders list.
Perhaps we should have.
BUT I still feel they would fall foul of the "don't compete with your customers" rule.
How could Microchip license microAptiv for use in microcontrollers successfully when it is sells so many microcontrollers?
How could Intel license high-end and mobile MIPS cores when it is trying to sell Atom and Ivy Bridge into the same sectors?
Intel+Marvell would be even better. Marvell is an ARM licensee and has several cores so Intel could quickly compete in the ARM market. IIRC, a chunk of Marvell used to be an Intel division, so that's another synergy. OTOH, it's possible that the ARM license wouldn't transfer to Intel, or that ARMH would make Intel pay dearly for it. Intel would have to use 25-33% of its cash to buy MRVL, however, so it probably won't happen.
Buying MIPS would just be pocket change for Intel, however. They have $15B in cash&investments, mostly short term, and MIPS market cap is only $0.353B. Even with a 40% takeover premium, the cost to Intel would be about 0.10 per share, less than half the quarterly dividend. Intel already has one of its best engineering teams in Israel, so there is synergy there as well. If Intel has any plans to acquire an ARM license, owning the MIPS IP could prevent ARMH from trying to extract too much from Intel. And Intel certainly has enough lawyers to monetize MIPS' patents and just reassign their engineers if they can't make a go of applying the MIPS architecture to mobile devices. But I very much expect that Intel could utilize the MIPS architecture for that as easy as they could do so with ARM. Especially with the latest Google developments.
But really, who wants MIPS? What has MIPS done for us or themselves lately (the past 5+ years)? And like all these suffering companies, what makes the next guy better at making it work than the last string of CEO's or owners? If you don't have real synergy, it's just going to keep withering. And how many other companies will MIPS take down with it when the architecture stalls (further)?
I'm guessing MIPS is way over-pricing themselves thinking they've got great value, when really they've just got a lot of companies by the short hairs. But they're 10 years over the hill. If nobody blinks, MIPS will be on Craigs List in another couple years.
I wouldn't put much stock in some super-secret computer-generated patent-rating algorithm (does it run on a MIPS processor?). I'd rather have two critical patents than 2000 frivolous ones.
I'm not seeing any big wins here, but potential for some big losses, with a domino effect to other IP companies. Tensilica, Ceva et al. maybe but MIPS may be too much woman for those boys to handle. Maybe ARM - or even Intel - buys MIPS just to milk them while snuffing them out.
Someone, please surprise me!
The Chinese want MIPS bad but the U.S. government won't allow it. My money is on Qualcomm, Synopsis, or Rambus. All of them are IP companies so it would make sense for them to acquire MIPS since it fits with business models. Qualcomm already uses MIPS and Synopsis would rather use MIPS than ARC because of the MIPS software ecosystem. The worst buyer would be Broadcom, that's the company which helped run MIPS into the ground.
Blog Doing Math in FPGAs Tom Burke 24 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...