Fundraising · Strategy

How do you start a silicon chip startup?

John Tobin CTO Limelight Mobile Inc. Owner

October 1st, 2015

I live in Silicon Valley, arguably the best place in the world to create a startup, but, the notion of a non-software silicon startup seems like an up-hilll battle on stormy seas, with a few tornadoes thrown in for good measure.  You can't exactly create an MVP so how do you do it?

Coincidentally, I do have an awesome idea for an augmentation to computer architect that would address a few security issues that many companies are experiencing. If you are interested then please drop me a line. 

Dave Lemley Consulting Technologist

October 1st, 2015

First, this is not my primary area of expertise, however I ws an EE in a former life, and I have recently worked (as a contractor) at a startup that made custom chips, so here is a couple things that might (hopefully) be useful.
1)  MVP; you might consider implementing your idea using FPGA first.  You won't need custom silicon to prove/develop the concept, and I imagine having this working prototype will make fundraising even easier.  As to whether this is viable depends on your product, of course.
2)  Things have certainly progressed since I was an EE, and there are now batch services where you can get chips made sharing a wafer with other folks designs.  This may or may not be feasible depending on what process you require.  You typically wind up with dice that you send to another facility that will do binding to the leadframe and encapsulation and testing.
hth

Craig Hunter Founder, Managing Partner at Element Technology Partners

October 1st, 2015

John - Have you tried Silicon Catalyst (sicatalyst.com)? Relatively recent incubator set up just for this space.

Steve Everhard All Things Startup

October 2nd, 2015

Jose perhaps not in foundries but definitely in silicon design, tools, testing and related infrastructure, with around a dozen last year. Soft Machines is a good example but there are others like Movidius and EPC who are operating in highly specialised fields. August Capital and Cadence, amongst others have a history of investing in this sector but with International Business Strategies estimating that it costs $132 million to design and test a typical chip using a 14 nanometer process the pressure is on the vendor to return around $1Bn in revenues to make it an attractive investment. That is a tough ask.

Steve Everhard All Things Startup

October 4th, 2015

Jose if your strategy for founding a startup is to pitch yourself into the midst of the funding activity then your odds of being funded are no better. I have no idea if Mr Tobin has a good idea or strategy for whatever his design idea is but there are likely more ways to skin the problem than rocking up to Sand Hill Road with a slide deck and a proposal for a new foundry. I suspect his major obstacle will be expertise to research his idea rather than funding per se.
Proving his idea will mean access to design tools and simulators which means selection of a cell library that is capable of hosting his design. A collaboration with a university might be a good step assuming he is willing to share IP and risk some publication. If he is talking security devices then there are additional physical and circuit design considerations to prevent perturbation and these skills are in limited supply. It's not the easiest startup but finding the right partners would be a first step to validating the idea, way before wondering what the funding landscape looks like.


Art Yerkes Computer Software Professional

October 1st, 2015

If you look at the recent stock price changes of several hardware companies, intel and sandisk come to mind, it's pretty clear that even the companies with the best lock-in and economies of scale in silicon production still must fight in this area.  azul, a company that used to make custom hardware to run java now produces a java vm for intel hardware.

One other way of creating an MVP is to add your functionality to an emulation environment, which also has the advantage of allowing you to develop the software support and customer interface before you ramp up hardware production.

Steve Everhard All Things Startup

October 1st, 2015

I'm not clear you are talking about custom silicon - are you? You are talking about the viability of HW vs SW startups as funding targets. The challenge for investors is always that startups based on HW often have higher startup costs, more complex distribution and issues like warranties and return policies that don't affect SW Having said all that there are a number of crowdfunded and VC backed solutions that demonstrate it is certainly possible.

Your MVP is obviously a prototype in whatever form you can construct it. Much depends on who the target market is and what it does and there isn't really enough information here to be able to advise you. Breadboard it if you have to in order to show working principles, and then produce an enclosure 3D printed with a clear plan of how to get the circuit inside -, which might well involve FPGA/FPLA or customised silicon depending on your application. In all probability you are likely to be able to get there with standard components unless security is your target area. Even then crypto chips are available with protocols for loading keys if that is part of your solution.

Obviously hardware has a ticket price and so the business model has fewer variants than an app or website. Much depends on whether software is downloadable/updatable or additional components are available to extend the purchase. Extended warranties and similar are upside purchases.

Zachariah J. Scott

October 1st, 2015

I am very interested in what you are working with. Could you please send me an email at: zscott@uikon.co ?

Jose Bohorquez Technology Entrepreneur

October 2nd, 2015

Steve: I don't mean to imply that there is zero chance of fundraising. But in answering the question "How do you start a silicon chip startup" and in the context of the rest of the question, I think it would be extremely hard to fund raise for the rough idea described.

I remember that as recently as 8-10 years ago, semiconductor startups were a common category for venture capital. Today, I'd be surprised if they make up more than a fraction of 1% of investments. The odds are not in your favor. Given the option, I certainly wouldn't start an IC company. It sounds like we're roughly in agreement. Cheers.

A. Andrew Chyne

October 3rd, 2015

John, starting a Hardware manufacturing startup is more difficult than starting a software startup. But if you have the passion for manufacturing, I am sure that you would succeed. Good luck John Tobin.

Jose Bohorquez Technology Entrepreneur

October 3rd, 2015

Interesting data here: https://www.pwcmoneytree.com/CurrentQuarter/ByIndustry