Read our International Electronics Forum 2008 News Roundup

Theo Claasen, vice president for business development at NXP, gave a masterclass on how to run a semiconductor company to the delegates at the International Electronics Forum 2008, organised by Future Horizons, in Dubai this morning.

Claasen kicked off with an anecdote about how he once presented to the board of Philips when NXP was still Philips Semiconductors. Presenting after him was a guy from Philips' lighting division. They both had foils showing their long term market volatility. "The foils looked similar, but the vertical axis on his foil showed a spread of +3 to -3 per cent, and on mine it was +50 to minus 50 per cent," said Claasen, adding, "that was one reason why Philips did not want to keep us in the company."

It's a fragmented industry, reckoned Claasen. A $256bn industry in which there are three companies with sales of over $10bn; more than ten companies  with sales of over $5bn; over 35 companies with sales of $1bn+; and 150 companies with sales below half a billion.

Against this background Claasen saw four trends: cost increase; de-verticalisation; specialisation and integration.

1. Costs: manufacturing costs are huge. A 65nm fab outputting 800k wafers per year costs $4-5bn. "So companies are building huge fabs without which they wouldn't be economic", said Claasen. Back-end manufacturing cost is increasing. It costs $500m a year to keep up in process R&D.
2. De-verticalisation:  "Almost all the system houses have spun off their IC divisions", said Claasen, "Siemens - Infineon; Motorola - Freescale; Lucent - Agere; Hitachi and Mitsubishi - Renesas; Rockwell - Conexant; and Philips - NXP, or they've sold, or joint ventured, their IC group to a semiconductor company e.g. Ericsson - Infineon; Nokia - STMicro; Alcatel - ST; Sony - Toshiba.
3. Specialisation: "We used to say you can't be in semiconductors if you're not in memories", said Claasen, pointing out that, since then, companies have been spinning out their memory divisions: Qimonda, Spansion, Numonyx, Elpida and Hynix. Some broad-based companies have become specialists by spinning off divisions. Others have always been in a niche, like Intel in microprocessors. "Intel has tried to diversify but now is mostly back to microprocessors", said Claasen.
4. Integration. "The specialists are finding that Moore's Law has hit them", said Claasen, "and function integration on a single chip makes it essential for these companies to merge with other companies to acquire knowledge of the whole  system, e.g: ST-Genesis; NXP-GloNav; AMD-ATI. There's more of this to come as many niche players are being offered for sale."

In formulating a successful business strategy, a company must take account of several rules: first, No.s 1 and 2 in a market sector make money, players 5,6 and 7 don't; second, scale matters; third, advanced CMOS manufacturing is a commodity; fourth, standard IP blocks provide no differentiation; fifth, system knowledge provide differentiation; sixth, there are ‘almost no' synergies between different application areas.

"It pays off to develop a strategy that takes this into account", said Claasen, "high market share means profitability. There's a strong correlation between being market leader and profitability."

In view of all of the above, the Claasen rules for success in this new industry environment is to focus on growth markets where you can partner with ‘market-shaping' players, and have a portfolio of business which split between; mature businesses, growth businesses and emerging businesses.

"Divide the R&D budget between these roughly 35 per cent for mature businesses, 55 per cent for growth businesses, and 10 per cent for emerging businesses", said Claasen.

At NXP this year, mature businesses will take 35 per cent of the R&D budget and account for 58 per cent of the sales; growth businesses will take 55 per cent of the R&D budget and account for 38 per cent of the sales; emerging businesses will account for 12 per cent of the R&D budget and account for 4 per cent of the sales.

And that's how to run a semiconductor business, the Claasen Way.

See also: Mannerisms, the blog of David Manners. Updated twice daily, it's the distinctive, entertaining, authoritative and never dull commentary on the semiconductor industry, from someone who knows. Sign up for the Mannerisms eNewsletter.