Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > EDA/IP

'Asset-lite' strategy keeps Moore's Law alive

Posted: 01 May 2007 ?? ?Print Version ?Bookmark and Share

Keywords:advanced CMOS process? Moore's Law and CMOS? paradigm shift in IC manufacturing?

Manocha: The only way to keep Moore's Law alive will be through collaboration.

The working alliances among big players in the IC industry have recently been shoved into the spotlight, as relationships within shifted and evolved into new forms. NXP Semiconductors' exit from the Crolles2 Alliance was one such move that shook the industry, and had analysts wondering what the next steps could be and whether the journey to advanced CMOS process technology!and Moore's Law!would stay on track.

Ajit Manocha, executive VP and chief manufacturing officer of NXP, believes that the widening gap between design productivity and Moore's Law will trigger a paradigm shift in IC manufacturing that will entail a breakthrough in the way companies work together. He spoke with EE Times-Asia about the manufacturing landscape of NXP and his view on how the industry is changing in relation to integrated device manufacturers (IDMs) and foundries.

EE Times-Asia: How would you assess the state of IC manufacturing? What is its impact on business relationships in the industry?
Ajit Manocha:Consolidation has started to happen. R&D cost is the biggest driver for innovation because it's becoming very expensive. Even the big players can't afford it, so there's a lot of merging being done for R&D. For example, NXP had to partner with Taiwan Semiconductor Manufacturing Co. Ltd for R&D, and we're a big player. So if that's happening for us, you can imagine how small players can't even begin to think about doing R&D themselves. Integration at the system level is another driver for consolidation because small players can no longer rely on standalone component plays. And the third driver is just how the big players swallow smaller players up.

In the next five years, most companies will not be able to afford a 300mm fab by themselves for 90nm and below, which costs roughly $4 billion these days. Also, the migration to bigger capacities is itself a big roadblock for companies to make sure that they can load the factories, otherwise investment becomes a waste of money.

Now, if you want to further complicate the picture, look at the back-end factories. Assembly and test companies have also been opportunistic like IDMs, because capital intensity is not very high in back-end, so most IDMs have decided to use their own capacity and go to subcontractors only when there was a desperate need. And when there was, it created very opportunistic behavior in subcons. Lack of stabilization resulted in more fragmentation in the back-end, so that's why they're not enjoying economic health.

The gating factor is product development, where design complexity is adding more barriers to the equation. R&D cost is going up with the addition of platform, software and IP development. This really needs to be tackled!otherwise, Moore's Law will not be able to continue.

I think the only way to keep Moore's Law alive will be through collaboration. Consolidation will provide communal skills, hence massive production and lower cost, closing the gap between design productivity and Moore's Law.

Regarding NXP's renewed partnership with TSMC, how has the relationship changed? How do you compare your partnership then and now?
We don't have the financial stake in TSMC!Philips has it. So NXP has been having purely business-type relationships with TSMC. And due to the foundry relationships it had with Philips Semiconductors before, TSMC also respects NXP because we provided them with a lot of technical strength and foundation for their growth. We have been doing the foundry business with them for years, and TSMC was also a joint development partner at the Crolles2 Alliance. We have a strong relationship with them!personally, business-wise and as founding partners. We have a lot of respect for the company.

Our partnership with TSMC in Singapore!the SSMC joint venture!has also been very successful.

If the industry is heading for consolidation, why has NXP chosen to go alone with R&D on advanced CMOS process technology?
In going forward, the biggest challenge is in R&D cost for advanced process technology. It's becoming so expensive that not a single IDM can afford it!and the middle tiers like us can't afford it, too.

Our partnership with Freescale Semiconductor has worked very well and run its course the way we wanted it to. However, in going forward below 65nm, R&D is becoming even more expensive. Meanwhile, companies like TSMC are doing it anyway because they are here to stay. Thus, we felt that it was better for us to partner with the leaders who are committed to this business and technology, and to developing next-generation nodes. It also really helps us to have development and manufacturing in the same factory!that saves us the cost of transferring from a Crolles2 fab to a manufacturing fab.

After our own announcement, Texas Instruments also announced that they will partner with a foundry for below 45nm. So it's not as if NXP has put forth an exotic idea!this is very sensible business thinking on our part. I think that other companies are following suit.

What do you think are the challenges that you'll be facing in going the asset-lite route?
I think there are more challenges for not being asset-lite. The real question is: When you have all these factories and you cannot load them, how do you get good return-on-investment on these fabs? Asset-lite makes it better with respect to managing production cycles, especially if we plan to grow our outsourcing to about 40 percent. That's what the asset-lite strategy entails!NXP will continue to milk and sweat our existing factories. All the business growth will go into further increasing our load through productivity improvements for small capital injections, while the additional growth will go into outsourcing.

The challenge is to grow our business to 40 percent outsourcing. If that can be achieved, it will be possible during down cycles to pull back some of the load from outsourcing companies so that they're not left high and dry, and we can still keep our own factories above the break-even point. I think that's the smart way of managing your assets.

How will your joint venture with Advanced Semiconductor Engineering Inc. (ASE) affect NXP's position in the regional market?
It only helps us in the regional marketplace. NXP has back-end factories in Thailand, the Philippines and Taiwan, and one small factory in China. In our efforts to penetrate China, ASE will inject capital to grow our China factory. That way, we can continue to serve our Asian markets from the Philippines, Thailand, Taiwan and China while we load our 100 percent-owned factories, and further expand China factories with capital injections from our partner ASE.

It's another example of smart asset-lite strategy!we are seeing the trend in the back-end as well, just as we did in the front-end with partners like TSMC. Basically, we are mimicking the asset-lite strategy in the back-end. It fits very well into our overall plan.

With the recent increase in NXP's operations in Asia, what will happen to your operations in Europe and elsewhere?
Let me make it clear: Our strategy is to really sweat our assets everywhere!whether it's in Europe or Asia. And part of the plan is to keep our assets because we're using an asset-lite strategy, not an asset-less strategy. Our objective is not to get rid of our assets in Asia!it is to really milk them and sweat them. If a particular factory becomes outdated because of portfolio changes or changes in business conditions, it's a business decision for us to decide on what will happen to it, whether it's in Europe or Asia. Whether a factory has to be upgraded, closed or whatever action needs to be taken is on a case-to-case basis. It's not a geographical decision!it's a strategic decision to sweat our factories, grow them if we can load them, and if we cannot load them, make a decision as any wise entrepreneur would do, regardless of the location.

The obvious hotspots for semiconductor manufacturing are China and Taiwan. Do you see any other Asian country on the map?
I think it's not easy to make that happen because Taiwan was very fortunate in that they were able to develop a lot of talent by importing from the United States and developing their own. China is catching up, but other countries in the Asian region will still have to work very hard to catch up on talent and infrastructure. So it's not going to be an easy task. Singapore has definitely done a very good job there, but if you're talking about other countries like India, Malaysia and Vietnam, I think it will take considerable effort for them to catch up with Taiwan.

- Selena Salang
EE Times-Asia

Article Comments - 'Asset-lite' strategy keeps Moore's ...
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top