A third emerging trend, said Synopsys’ Pan, will be more M&A among China fablesses.
noted that IPOs in both Shanghai and Shenzhen markets have been
suspended over the last year or so. VCs who seeded those startups,
sooner or later, “need to find an exit,” Pan explained. This gives extra
impetus for mergers or acquisitions. Consolidations among Chinese
fablesses will pick up.
Meanwhile, Western companies are more
likely to snatch up “a team” [of engineers] in China, instead of going
through M&A, Pan added. Such smaller deals often remain invisible,
because the companies find no reason to disclose them.
contrast, will any China fabless be interested in buying baseband
technologies from Ericsson (i.e. after ST-Ericsson dissolved) or Renesas
Mobile? Pan thinks that’s unlikely. Those multinationals were initially
interested in getting bought in their entirety. In order to attract
China fabless, they need a good technology licensing program, Pan added.
Pan summed up the state of the Chinese electronics industry today as follows.
China’s electronics industry--including everyone from fabless to foundries--finds itself trapped in a perpetual quandary.
one hand many Chinese fabless companies are so close to the market
[design houses, system OEMs/ODMs] that they’re under constant pressure
to respond to their customers immediately. “The push from the market
keeps them very aggressive,” and makes them yearn for advanced process
nodes all the more, he explained.
On the other hand, many
Chinese electronics firms lack experience. They lack IPs. Asked what
roles companies like Synopsys play here, Pan said, “We try to fill in
the gap.” Cadence’s recent announcement on Tensilica acquisition aside,
getting ready for “complete solutions” is one of the key trends Synopsys
pioneered. “Last year alone, we pulled off nine acquisitions including
Magma Design Automation, SpringSoft and Eve.”
Chinese fabless in
designing next chips, for example, would want all the connectivity IPs
around ARM. “When you look at Synopsys resources vs. revenue ratio,
you’ll find that ratio is higher” in China, he added. It’s because EDA
companies need to provide more help.
Companies are doing the baseband processors because that is the easy part. Buy an Arm core, which is what everyone wants to use, integrate some peripherals and you can offer something. The major companies are doing their own apps processors but they are having problems with the modems. The hard part is the complexity of the modem and the software protocol stacks that go with it. You can find companies that will develop a protocol stack for you but don't expect it to work as advertised.
I think many of these baseband IC companies you are spotlighting, plan to get some hardware ICs out there and hope somebody writes some software to go with it. Good luck with that. You're going to be staring at your pretty hardware ICs for a couple of years.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.