Chinese language news sources are reporting that the Institute of Computing Technologies of the Chinese Academy of Sciences is considering plans to buy a 20% stake in MIPS technologies, the California-based semiconductor IP development firm founded more than 20 years ago by John Hennessy, who is now president of Stanford University.
The sale is significant for two reasons. The first is that MIPS is a storied company, whose original chip designs, while now used primarily in embedded devices, once powered everything from desktops to supercomputers, and are still used as teaching platforms in many electrical engineering programs. The second, more important reason this sale matters is that China’s home-grown Loongson processor runs an extended version of the MIPS architecture, and the Institute of Computing Technologies (ICT) has a full architecture license that allows its engineers to both build MIPS-powered chips and to extend the architecture itself with new instructions. This has allowed the ICT to develop successively faster generations of Loongson processors, from the first prototype to the Loongson 2F, which is currently available in netbooks and low-power desktops from Chinese manufacturer Lemote. The third generation of the Loongson processor might even be fast enough to allow China to build the world’s fastest supercomputer using Loongson Chips.
ICT is following in the footsteps of countless tech companies before it (e.g. Apple and Imagination Technologies) by acquiring a stake in a company that makes intellectual property that is essential to its products. MIPS has so far declined to comment on this potential acquisition.
MIPS is a small company and does not manufacture chips. Instead, like its competitor ARM, MIPS maintains, enhances, and licenses the existing MIPS architecture. If I’m reading its quarterly reports correctly, it has a market capitalization well south of $100 million, which means that acquiring a 20% stake in it would be trivial for the cash-rich Chinese government, though perhaps quite significant for the ICT and the Chinese Academy of Sciences, which no doubt have their own budgets.
MIPS’ architecture has long been the only portion of the IP behind the Loongson processor that wasn’t entirely invented and controlled by organizations within China–potentially a sore point for government backers of a processor that seems to be as much about national pride as advancements in the field of semiconductors. By buying a piece of MIPS, the ICT has taken a small step toward addressing that issue.