“Silicon Valley is a misnomer” Rick Clucas told me, “there are no silicon companies in the valley anymore, apart from the old timers. The reason silicon start-ups fail so often – and they do – is that the funding is so difficult. Software start-ups and the new generation of dot coms find it a lot easier to attract investment. And that’s where the Silicon Valley VCs put their money. Not in silicon.”

We were sitting at a table in Rules, a restaurant in the Covent Garden area of London. Rules is truly unique, like Nelson’s Column or The Tate Modern. It’s a genuine English restaurant, the food it serves is genuinely English and the waitresses dress like English maids. The wine however, is not English, which is no cause for complaint.

Rick is the CEO of a silicon start-up, called Coresonic, whose technology I’ll discuss in a different posting in a few days. His previous silicon start-up, Ignios, failed through a drought of funding and that’s what he was bemoaning. The company had highly promising technology. Basically, it was a system-on-a-chip. It was multicore, but didn’t require any kind of parallel programming. It did the parallellizing itself on the chip, without much loss of efficiency.

Now I don’t know the technical issues in depth with this company, but think on this: The company was founded in 2003 when no-one was talking openly about multicore or the necessity for multicore. It wasn’t yet common knowledge that Moore’s Law was finally being violated (and not in a good way). So the designers of this chip were on the money, both in respect of it being a sensible development and in respect of timing.

It didn’t matter though. The company took too long to get a committed customer and the VCs pulled the plug. No other VCs were willing to step in. It’s a familiar story, and it should be, because that’s what happens to 80% of start-ups.

But silicon start-ups have it worse. According to Rick, silicon start-ups typically take 3 years to get to a demonstrable prototype and then it’s likely to take another two years to get a customer signed up. So that’s “5 years to cash” and the investment required will be around $5 million at current prices. Customers, such as mobile phone companies, car manufacturers or consumer electronics companies do not lightly use new chips. There are not many sales opportunities and the incumbent chip vendors are difficult to dislodge, no matter what technical advantages your silicon may have.

It should be no surprise that there are not many stellar start-up stories in the chip market. Even if you manage to break in, as ARM did, it takes years to get to volume. Contrast this with Skype or YouTube, or MySpace or Facebook. Such companies may be rare, but the time to value is short and the achievable value for the VC is very high. It’s simple mathematics really.

That’s why the silicon start-ups have a rough ride.

The list was distilled from conversations and presentations at IBM’s EMEA analyst conference; Analyst Insights 07.

1. IBM’s Cell chip has become a rapid success. It’s the engine of Sony’s PS3 and it is also in use in a variety of commercial areas, including health devices, consumer electronics and specialist devices in the military, aerospace, etc.
2. IBM’s Power chip goes from strength to strength. In the games market it is the competitor to the Cell chip – the CPU for the XBox and the Nintendo Wii. It’s also one of the reasons that IBM’s pSeries business is going gangbusters.
3. Not wanting to harp on about the Power chip, but IBM is projecting that it will produce a 10 petaflops computer by 2010. That’s about 10x faster than today’s fastest bad ass computer (What’s a petaflop? It’s what a teraflop becomes when it grows up).

Add it all up and you have to conclude that IBM is fiercely competitive in most of its business lines.

4. IBM does engineering consultancy. It is behind many devices from many electronics brand names, having collaborated in building everything from high definition TVs to mobile phones. This is no small business for IBM
5. IBM is ahead in virtualization. It is the first computer server company to produce a complete virtual architecture. Particularly impressive, and I believe useful is its virtual storage.You want virtual? IBM’s got it.
6. IBM does blades better than the competition right now. In it’s Bladecenter, everything; disk, tape, etc. can be plugged in as a blade.
7. The mainframe market continues to grow. IBM is now pushing the mainframe as the center piece of a consolidation strategy – and demonstrating what it can achieve by consolidating 3900 of its own Unix servers onto 30 mainframes. (Apart from the floor-space gained, you get performance improvements, management economies and reduced heating problems).
8. IBM knows SOA. It does SOA comprehensively from the software (it has a very comprehensive stack) through to the build (it has many consultancy engagements). And by the way, SOA is now mainstream.
9. IBM is committed to cool. As far as data center cooling is concerned, it is in the game at every level from the chip through the server devices to heat flows in computer rooms. (Nowadays this is obligatory for a server vendor, but IBM competes strongly in this area)
10. IBM claims that “Big blue is big green”. But it’s not simply hype to convince the world that IBM isn’t melting the poles and killing polar bears. IBM now has specialist consultancy activity that can help a company to reduce its carbon foot-print – and it’s comprehensive. It covers every aspect of “green” from energy efficient property to the organizational impact of telecommuting.

Add it all up and you have to conclude that IBM is fiercely competitive in most of its business lines. It doesn’t look like much of a dinosaur to me.

It’s a good question. $15bn is about one tenth the value of IBM and that has to be a stretch (if not an absurdity) given that Facebook’s revenues are counted only in the tens of millions. Nevertheless, Microsoft accepted such a valuation and showered Facebook with $250 million for a 1.6% share, just to prove it. Other investors threw in $500 at the same valuation, so it isn’t only Microsoft that will buy at that “remarkable” price.

You could argue that “if Facebook is worth $15bn then MySpace, which was snapped up by Rupert Murdoch for a little over half a billion, is worth more” – because MySpace currently has a larger user base. The problem with that idea, is that MySpace is losing customers to Facebook at a significant rate – and for all that anyone knows, it may go into terminal decline.

The $15bn is a curiously round number – as though someone has said “it must be worth more than $10bn, so it has to be $15bn, right?” Irrational exuberance only works in units of $5bn.

Imho, there is no doubt that Facebook is valuable property, because it has become an anchor site on the web where people of all ages go regularly. You could claim that it was “sticky”, in the old dot com parlance, but it’s more than sticky, it’s your personal web presence. More than that, it now has a developer community developing facebook “widgets” to increase the possibilities for every facebook user. And that’s fine, but can it all be transformed into gold?

Well it might happen, if you can make Facebook even more of an anchor, by turning it into “your primary email site” and “your primary chat capability” and “your search capability”, etc. Oh and by the way, you’ll need to do something very slick in the area of advertising.

But that’s where the $15bn comes from, because some commentators, and we presume, some investors, believe that all this is possible. It’ll be a neat trick to pull off, but I’m not sure challenging Google is do-able. Google has become a tar-baby, whoever and whatever touches it sticks to it. You need a lot more than Facebook in its current state to undo that.

Facebook is a site to watch. It innovated its way past MySpace and it may have a stellar future. Despite its meagre revenues I believe a valuation at $2bn to $3bn would be justified. $15bn is irrational exuberance, in my book.