Electronica 2008 - Read our full show coverage from Munich 

Talk to European venture capitalists and you will find very few show interest let alone enthusiasm for fabless semiconductor start-ups.

Too expensive, too long to revenue, too few high volume opportunities, too much competition, the list goes on - but is it really like that?

There are two big changes in the industry that I suspect have driven this view: The cost of developing complex SoC’s on deep submicron process and the cost of developing application software for complex SoC’s.

It costs more, takes longer, with fewer markets big enough to offer a return on investment.  Everyone is chasing the same market, so it really is like that.
QED?  End of article?  Go dream up a Web2.0 social networking start-up? I think not.

Driving this view is the assumption that competitive products require leading edge technologies. If you can build a competitive product with a mature or ‘mainstream’, therefore cheaper, semiconductor process then, with a few changes on the go-to-market strategy, and extra investment in application software, the fabless business model still works well.

We developed a 16-bit flash microcontroller. Not original, certainly not the first. Arguably it‘s the last thing the world needs and for a time even the company struggled to accept that fact. We feel strongly that it’s the best 16-bit MCU - but again that’s missing the point.

We developed an MCU that when designed into your system will save you money - and that’s what the fabless semiconductor business has always been about.

The fabless model that worked used state of the art semiconductor technology to integrate more onto a SoC, replacing multiple devices with one single device and saving the customer money. This simple model drove the Fabless industry, providing opportunities for new entrants right up to the point where cost and complexity of digital SoC’s became a problem.

So how do you build a competitive product with a mature cheap semiconductor process?  It’s not easy… you have to innovate… get more in for less cost and power … and add speed. That’s good silicon design and lots of people do that so it’s also not enough.

You must save your customer money.  That means finding things that you can displace and reduce the bill of materials.

The fabless model takes a holistic system level view, stepping up a level and delivering products that have a predefined function, are easy to use and save money. If the market can’t afford the technology to integrate everything monolithically, then integrate onto a production ready PCB and make it easy for the customer to access those savings.

It’s the same model as system-in-package but more flexible, at least in our implementation with production ready reference designs for the user to modify.

If you can build a competitive product with a mature cheaper semiconductor process then, with a few changes on the go-to-market strategy and some extra investment in application software, the Fabless business model still really works.

Kenn Lamb is CEO of fabless microcontroller supplier Cyan

Full, extensive editorial coverage will be provided on Electronics Weekly, in a special Electronica section. Interviews, latest releases, details of the show, such as floorplans, exposition of latest technologies... Simply bookmark: www.electronicsweekly.com/electronica2008

Sign-up for the ElectronicsWeekly.com newsletters
Daily Latest (Daily)	Weekly Roundup (Weekly)	Mannerisms (Weekly)	Circuits (Fortnightly)	Made By Monkeys (Fortnightly)