I would agree with the others that VC is not interested in Fabless startups and large SoC based companies. What I would suggest is that the sift should be into the IP development and the small to medium sized startups; this is where there is real potential for ROI and the most opportunity. The steady migration of IC fabrication and design/development off shore is going to hurt the US in terms of competitiveness and balance of trade. We must be innovators and creators of the next technology or the US will continue to fade in importance. I suggest that we have plenty of great ideas/people to fuel startups, what is lacking is VC and the willingness to risk. How do we get the ball rolling? Lets not look to the past but to the future.
Regarding complex SoCs from VC-funded fabless startups, I agree -- the game is over. The risk vs. reward is simply too unfavorable.
An SoC development in the latest CMOS technology is a very large bet, and only an organization with a broad portfolio of products & sales in multiple markets can afford to make -- and possibly lose -- such a bet.
On the other hand, not every IC development is an SoC that is built with the latest bleeding-edge process. I think there are still opportunities for VC-funded fabless startups that have unique IP that can be developed in more mature (cheaper) processes -- a more attractive risk-reward profile.
@Tamza; Responding to your final comment "Time to take on what Grove and others are now saying (to 'protect' their legacy?): BRING BACK manufacturing to the US." Is it not too late by now to reverse the 'runaway' transfer of the US design-and manufacturing technology to Asian countries specifically China (ROC)? It's gone or challenged like many other of the 'former' mighty US industries (Steel, textile, automotive, machine tools, commercial aircraft etc.), thanks to the free trade, outsourcing and offshoring polices in the name of the global economy and most of all higher profits, profits, profits for ???.. !
Agree with @Tamza, it has been difficult to make money on semis investments for several years at least, and there are many VC funds that will not touch fabless IC companies...IC design cost is too high compared to revenue potential it can generate...a few ICs needed for PCs, cell phones etc. will be designed by Intel, Nokia, etc...the game is over! Kris
The value of the semiconductor industry as an investment has been in question for decades .. I recall a conversation from 1980 or so, "We are losing on this product, but we will make it up in volume." Now that principle worked fine when there were HUGE volumes so that eventually the cost came down enough and prices did not drop as much from introductory levels (ie there was a cross over) that in 3-5 years a respectable ROI could be made. The shorter life cycles make this difficult, if not impossible. I did a study in 1982 of the ROI of the pure-play semiconductor companies and as an aggregate the industry had LOST money, ie: NEGATIVE ROI; and if you remove the almost-monopoly Intel things were really bad. The main justification to continue, as stated by Jerry Sanders, was that it was (even is) the OIL of electronics, even the national economy. So now what we have is investment (or lack of it) in the US, and returns in China (high volumes production, benefit of lower learning-curve costs, etc).
Time to take on what Grove and others are now saying (to 'protect' their legacy?): BRING BACK manufacturing to the US.
Actually, Warren, your commentary supports the position of the authors. Save for Achronix, which has VC funding, the rest of the activity that you cite is by mature corporations. VC funding has been turning away from fabless semi and EDA, Web 2.0 is what most of them are now chasing. VC returns are also looking poorer as the dot.com excesses are now leaving the 10-year window on returns.
It should be recognized that systems are inherently tailored for a specific set of functions, not a simple description like memory or microprocessor. Hence, they do not fit the classic high volume manufacturing model of those commodity products. So a SoC product cannot be high-volume. It is same with SiP. Silicon cannot sustain innovation with cost being a key concern.
Considering Figure 2 the authors, I feel, fail to give enough credit for the sad state of VC affairs to the dotcom crash and to the US Government via Sarbanes-Oxley. And the implication that the current state of VC activity implies impending marginalization seems way, WAY off the mark to me given the flurry of activity and innovation going on within the (revenue generating) semiconductor industry right now; see Xilinx (Stacked Silicon Interconnect), Intel (Achronix), Samsung (their entire LSI Division), AMD (Fusion), GlobalFoundries (their very existence), ....
Blog Doing Math in FPGAs Tom Burke 15 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 ...