Need of Company Sponsored VC Fund in Semiconductor

[barun]

We all are concerned about the declining funding of new semiconductor start-up which is a severe concern for driving innovation in semiconductor industry. But the number of M&A deals happening in semiconductor industry is still maintaining healthy rate. This indicates the demand of innovation or new technology is still high in the industry.

Hence the question arises what can be the solution. Without availability of fund the rate of new product launch will decrease. But if we look more accurately it is not the lack of availability of fund which is preventing VCs to invest in semiconductor industry, it is the perceived risk associated with the investment which is the main concern to all VCs. The semiconductor industry has become very complex in terms of technology and deciphering the value behind a new innovation becomes a challenging task to the VCs communities.

Perhaps the solutions may be creation of dedicated funds by semiconductor giants which can be invested into new start-ups. If the semiconductor companies invest in start-ups operating in their domain then they will be much better equipped to understand the value behind the new product. This will automatically reduce the risk of investment. Moreover these companies can play a role of strategic investor where the investor can bring help to promote the product among potential customers, augment the product to complementary products etc. This will definitely increase the success of new start-ups.

<script src="//platform.linkedin.com/in.js" type="text/javascript">
lang: en_US
</script>
<script type="IN/Share" data-counter="right"></script>

 

[Daniel Payne]

FYI - Intel Capital has invested in:

• 11 fabless semi companies
• 27 semiconductor related companies

 

[Daniel Nenni]

This is part of what I do for my day job, I work with emerging EDA and IP companies and funding is always an issue. Companies like Qualcomm, Altera, Xilinx, Broadcom, Apple, etc... do invest in emerging technology companies. They do it as JDAs or joint development efforts investing via product purchases. This not only provides capital, it provides product direction and inspiration for innovation. TSMC also works with emerging technology companies and invests in similar ways: Product purchases, free shuttles, test wafers, etc... EDA companies do the same offering free or discounted tools for equity. Times have certainly changed but clever people still find a way to get funded.

Given the SoC revoution, IP companies are springing up everywhere and that is where I see the most funding activity in the semiconductor industry. Most of these companies are self funded with money from the founders previous exit or from their 401k plan. Free tools, free shuttles, and JDAs with the top fabless semiconductor companies can go a long way.

 

[barun]

Hi Daniels,

Thanks for your valuable replies. Actually I wanted to mean that funding in semiconductor industry is changing the direction where the company sponsored funds are replacing the traditional VCs. Yes, it is happening already (my heading or structure of writing was not right).

But to accelerate the process we perhaps need more than that. Perhaps we need a collective funds (better to say investment consortium) created by an EDA company, a foundry, an SoC company, a test and measurement company which will invest (in terms of free tools, free shuttles, free integration with complimentary IPs, free characterization and certification) in design IP companies in exchange of equity. The formation of such collective funds will reduce a lot of effort in the due diligence and negotiation process and makes the cost of fund lower.

Regards,
Barun

 

[Pawan Fangaria]

Looking from the business and holistic angle what we see is that innovation does happen, followed by M&A. Yes, VCs need to look at from that perspective on how promising a technology is which is right to invest in and can provide big returns. The same can be done by those who acquire a start up after its success, provided they can visualize the pro and con as the VCs do. And my belief is that they (the companies like QCom, Apple...) can do this with more ease because they have customer insight (long pending giant requests - i can name in EDA space - analog automation still is an area to invest) on needs and future directions. What is more important is that an analysis like "what if I invest in this particular technology from now till 3 years? Will the overall investment be much less than what I would incur in acquiring the same from open market 3 years down the line" is needed. However, I have not even seen that kind of reverse analysis from acquisitions which have already happened so far, that is "I acquired it for $300M. I know it took 3 years for them to develop. What if we would have started it on our own 3 years ago? What could have been the benefit?" I think these kind of analyses can bring more confidence in investing.

Having said that, the idea brought up by Barun that a couple of important players (such as EDA, Design, Manufacturing companies) in the chain should join together to create that fund. True, that's a nice idea and can become practical once the confidence is developed. With the success of a few, more and more players will feel at ease to join.

 

[LinkedIn]

Dr. Leonid Shvartsman
It just seems that venture technology in silicon technology are a big risk. Recently, in different places have accumulated so many new technical solutions, that the establishment of the experimental chain of the new technology will help to minimize the risks. And it's worth it all no more than $ 5 million Production cost will decrease by 2 times, capital costs - a factor of 2, the construction period of 2 times. And if you apply local projects, it is still faster. Everyone is used to make polysilicon through MG-Si. But there are also other areas.

 

[barun]

A couple of important players (such as EDA, Design, Manufacturing companies) in the chain should join together to create that fund

If we look in good amount of cases these fund can be created without committing real money. An EDA company can say "I will keep x number of licenses allocated to be given to new start-ups against equity instead of selling in the market". A foundry can say "I will allocate x% of my MPW slots to start-ups against equity" and so on.

Also there is also other ways of collaboration between the investor and the start-up. An EDA company can give free tool licenses to an IP company against joint rights on the IPs. An foundry can give free MPW slots against preferential pricing. An SoC company can integrate IP from a start-up in their new design as trial and can ask for preferential pricing if it is actually used in the SoC. A test and measurement equipment company can collaborate with an IP company where their test software for new standard is validated with the test chip of the IP.

In this ways a new start-up can be promoted without a real transaction of money

Regards,
Barun

 

[Pawan Fangaria]

That's a good way of looking at it. However, opportunity cost of those licenses and services has to be looked into. Also, start up needs stable monitoring funding as well for several other expenses such as staff cost, facility management etc. And another big question is how much the management is willing to dilute the equity in that case.

 

[LinkedIn]

peter gasperini
It won't work, it's just substituting companies for VC's. The entire methodology of chip start-ups is broken.

Start here (it's in three parts.) This is the way forward to revitalize silicon valley.
http://www.eejournal.com/archives/articles/20120426-siliconvalley/

 

[LinkedIn]

peter gasperini
IP is part of the infrastructure for chips - the only truly bright spot in the chip industry, and it's seeing lots of value transfer coming its way. If you want to build actual chips, though, you have to do it differently from the Jeffrey Moore model (Inside the Tornado et al.) Read the link through, Daniel - there's more detail in the link than I've provided here, and it'll become apparent as you go thru it. IP, tools & methodology for front and back end, market penetration, revenue ramp, etc...it's all there. I think you'll find big chunks of it resonate strongly with what you do and you'll enjoy the overview.

 

[barun]

Start here (it's in three parts.) This is the way forward to revitalize silicon valley.
http://www.eejournal.com/archives/articles/20120426-siliconvalley/

Really good articles. It shows some valid ways where chip start-ups can save cost. I agree few of them

1. Backend activities (RTL/ Netlist to GDSII) is a well established process now with clear input/ output definition. Rarely a SoC (perhaps except the processor ones) gets differentiated in the ways physical implementation is done. Hence there is no point for a start-up to invest in costly EDA tools and also flow development. It is always better to outsource the work to vendor who has already established flow and can amortize the EDA tool cost across multiple customers.

To ensure desired outcome of the physical design process, start-up can try to negotiate outcome based pricing with the vendor. The pricing depends on the area, power achieved

2. Some of the marketing activities. If the potential customers of SoC (for power management chip, data converter) is small companies and are in large numbers, importance of marketing through social media, through writing articles in leading journals, through creating webinars and posting in youtube. But this method may work if the customer base small and large companies (for example customers of storage SoC).

But there is further area of improvement

1. Verification (though not like physical design) has also becomes quite automated and it is better to outsource the activity particularly if chip has lot of standard interfaces. The vendor can provide people who are experts in those standards for the required duration (typically for the development of verification suite) which is small and does not justified hiring

2. Even the manufacturing cost can be reduced by using third party vendor. 28 nm is not always needed and there may be very minimal saving in die area (and hence die cost) compared to higher nodes, particularly if the SoC is IO heavy and have lot of analog components. A vendor can suggest proper technology nodes from his experience. Also he can suggest right package, tester and number of test sites

But most importantly, I feel the innovation in semiconductor industry is too much focused on consumer and communication domain which is too saturated. There are other areas like medical, industrial, automotive where innovative SoC is needed. The innovation needs to come from RF, analog, MEMS areas. The investment required in these areas are significantly low

 

[Paul McLellan]

GSA actually had an afternoon on just this topic in July. I wrote about it on SemiWiki at the time here. Some of the players mentioned in this thread such as Intel Capital were in the panel sessions. SK Telecom have been setting up a fund. They were on the panel and I interviewed Angel here. A theme running through much of these sort of discussions is the need to be much more capital efficient. For EDA companies where you can (usually) get a software product to market without requiring an enormous team, that is a lot easier than for a company doing an SoC design which requires a major investment.

But there is another reason that the investment is broken. The exits are not there to make these investments attractive. In many cases the markets are not there -- how many markets have enough unit demand to justify a 20nm design. Mobile, of course, but since Apple, Qualcomm and Samsung design their own chips, it is hard to sell into that market profitably.

 

[John Swan]

@Barun, the partnership of companies does happen from time to time but in a different manner. But the first thing that came to mind was companies that are individually funding startups, already commented on.

As far as pooling money for VC, I don't see them ever forming a separate VC company that then goes out to look for startups to invest in. Rather they pool their efforts on a case-by-case basis. In the early 1980's companies partnered to create the Microelectronic and Computer Technology Corp (MCC) for which I was a technical liaison for some time. MIPS was formed initially as a consortium. Motorola got several companies to share the risk on the Iridium satellite communication system by pooling investment. I think the efforts will reamin more focused than I think you would like to see mostly because of both the high investment level and risk involved in this industry. Does this make sense?

 

[barun]

how many markets have enough unit demand to justify a 20nm design. Mobile, of course, but since Apple, Qualcomm and Samsung design their own chips, it is hard to sell into that market profitably.

Hi Paul,

I think this is common mistake start-up makes. We are too much obsessed with the cutting edge technology nodes and think the new SoC has to be in that technology node. There are other technology where investment is significantly lower like SiP technology. If the SoC has huge memory or RF/ analog blocks it may be better to have separate dies for digital, memory and RF/ analog

Regards,
Barun

 

[Staf_Verhaegen]

Rarely a SoC (perhaps except the processor ones) gets differentiated in the ways physical implementation is done.

I think this assumption breaks down if you are doing low-power design. Here experience of back-end with low-power techniques will be differentiating factor as well as the interaction between the chip architecture and the back-end.