RIP Microprocessor Startups

6 Comments

I’ve been talking about the enormous amount of cash it takes to create any kind of chip company and expressing doubts about the number of startups we will see getting financial backing to create truly innovative ideas in semiconductors. Analyst Linley Gwennap apparently feels the same way, because he looked at the sale of P.A. Semi to Apple and the recent sale of Montalvo to Sun (likely for less than the $73 million it raised) and concludes:

“The Apple deal will double the $126 million invested in P.A. Semi, a positive return but modest by VC standards. With the possible exception of RMI (which predates Dobberpuhl’s company), we expect P.A. Semi will be the last processor startup to generate a positive exit after such sizable funding. Montalvo will probably be the last processor startup to even raise that kind of money. Microprocessors have become a big-boy game; newcomers need not apply.”

As the last big microprocessor startup standing, Raza Microelectronics (RMI) was the brainchild of Atiq Raza, who formed a company that was later bought by AMD and turned into one of the company’s core microprocessors. RMI has raised more than $120 million to build communications and networking processors. I don’t want to believe it’s the end of startups trying their hand against the likes of AMD or Intel, but until we come to a breakthrough in materials, ways to reduce the IP hurdles or the cost of masks and design, entrepreneurial chip engineers will have to focus on power managment and cooling, MEMS and RF.

6 Comments

Tom Lynch

Bosh, there are at least a half dozen processor start ups right now. There will always be new processors that do things differently to deliver value and investors who make money from them. Access to fabs will continue, and new ways will be found to aggregate chips for mask sets and to share the cost.

People who do not envision the future are always predicting it will be just like _right now_. Incumbents in office, or in industry, always advertise that experience (entrenchment) is necessary.

Also, what might be confusing things here is that we just witnessed the field narrow for the x86 clone chip battle. However, up and coming is the struggle for the media convergent processor, especially at low power. Market demographics are also shifting. So what you see in your back yard may not be the most important story anymore.

-tom

Kevin Krewell

While Linley has a valid point, I think there’s still opportunities for processor startups – they just need a different business plan than what we’ve seen recently. Too many of these companies are focused on business models that are based on winning Cisco designs (Sibyte, Raza, Cavium and PA Semi) or the old “if we only get 1% (or 2% or 5%) of the PC market we’ll be sucessfull (NexGen, Transmeta, and Montalvo).

An example of a company that thought different is Luminary Micro (Austin). They are using mature (cheap) process to make Arm-based microcontrollers. It’s not as sexy as multicore, multi-GigaHertz state-of-the-art processors, but it fills a need, not an ego.

Comments are closed.