At least two on the panel of
five VCs said they are and will continue to invest selectively in
semiconductors. Ketan Patel, principal of New Venture Partners (Murray
Hill, N.J.), said he has at least four chip startups, two in stealth
mode, working in areas such as non-volatile memories and switch fabrics.
highest flying IPOs pick one specific component not available as a
merchant part, raise the bar in that area and are rewarded,” said Rust.
He gave Palo Alto Networks and Fortinet as examples of companies that
made “judicious selections of specific components where functionality
deserved to be implemented in a custom design.”
for investing in semis is the multiples the sector is rewarded with are
not as attractive as they were, but I am absolutely still looking a
semis,” he said, adding “USVP has had more semi IPOs to date than anyone
Panelists debated the opportunities SoCs present for
server microprocessors. A Microsoft data center manager in the audience
said top-of-rack switches could disappear in three to five years as
networking is sucked into integrated CPUs.
“Convergence in the
data center has many aspects,” said Asheem Chanda, partner at Greylock
Partners (Menlo Park, Calif.). “In the last two years, server and
storage products have been combined…but it’s still early to see storage,
compute and networking in [server] SoCs,” he added.
noted storage, computer and networking technologies change at different
rates and may not be suitable for integration in some apps. “What
happens to Broadcom or Marvell when some of networking gets absorbed
into CPUs,” asked Rust, noting the possibilities for disruption. Related stories:
On the surface, this is good news for the chip industry. But I hope the VCs who pledged their interest in backing semiconductor firms aren't just playing lip service to the industry. They've been telling us over and over it's just not worth their bet.
It's nice that the VCs are willing to look at chip technology but what about chipmaking equipment technology? The Bay Area VC community fueled much of the system, sub-system and critical component technology developments in the 70's,80's and into the 90's but then they turned tail and ran. As a result equipment technology is stagnating and innovative products are left stranded after entrepreneurial resources are depleted. This does not bode well for the future.
Great point, but with fabs consolidating into fewer hands and critical items like EUV becoming big industry pushes, is there a market for a startup there?
I recall sometime ago Sun co-founder turned investor Bill Joy told me there are opportunities for startups leveraging the infrastructure of what you can do with all that wonderful installed high tech gear.
I was sitting next to Rick when he asked the panelists if any of them were investing in semiconductor companies. Only 2 of them said yes, with only 1 start-up named. That was hardly a great endorsement for IC start-ups, IMHO.
At other VC panels I've attended, there was a great reluctance to invest in IC start-ups due to high capital expenditures required and a relatively long time to gain ROI via exit.
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.