With wireless becoming the leading mantra at the chip division of the Dutch electronics conglomerate, McGregor's company is working on projects to insert wireless functionality into clothes, cars, books, plane tickets, TVs, keyboards and homes.
Despite the seemingly boundless opportunities, making money won't be easy. Semiconductor manufacturers will increasingly have to come up with cost-cutting strategies to survive in an era where products will sell for pennies and the cost of building factories to make them will run into the billions.
McGregor, who helped design the interface for the Xerox Star, the first commercial computer for office use, also was the chief architect reporting directly to Bill Gates on Windows 1.0 during his time at Microsoft. He recently spoke with CNET News.com about his ambitions for Philips in this fast-evolving future.
Q: Can you give an overview of Philips Semiconductor and how it fits into the overall company?
A: Philips is a fairly broad company in the U.S. Go to any Home Depot story and Philips is the one there for lighting. Of course, we spend most of the television advertising in the consumer-electronics area. Unless you were in a cave, you probably noticed our flat TV ads. Semiconductors are an important part of the whole. We probably represent 15 percent of revenue.
We do about 11 percent of our sales internally to the company, while the rest of it comes from outside. We did $4.4 billion in 2001. We were unprofitable last year, as were most companies in our industry.
What does the semi division specialize in?
We're quite diversified, but we specialize in connected consumer solutions. More than half the TV sets in the world have our chips in them. We have a top-three position in cellular phones. We are now number No. 1 in car-infotainment.
We laugh at it now, but you could have a kiosk so when you walk by it says, 'Congratulations for buying those shoes.'
Are there any good markets right now for chips?
Automotive remains fairly strong. That tends to be a fairly resilient market. The cell phone business has been relatively soft, although it's a relatively large market. But within each market we see niches that are doing well.
In cell phones, for instance, one feature people are adding is FM radios in cell phones. Pretty soon, all reasonably priced cell phones will have FM radios in them, and right now we're the only ones shipping in that space. Another thing that is popular in the cell phone space is Bluetooth headsets.
Those are finally getting popular?
It's interesting, because people play with them and they are actually kind of cool. Another thing that helps is that a number of countries and states in the U.S. have gone to a hand-free requirement in cars. Bluetooth, either as a headset connected to the phone or built into the car itself with a way to hook into the audio, will become popular. We're now shipping approximately 650,000 to 700,000 Bluetooth units a month, so that business is beginning to pick up.
802.11 is absolutely hot. Last year we had 50 percent market share in radios for 802.11. There is a new standard coming out called Zigbee that is a low-data rate. It is 100 kilobit per second. The nice property is that it will run on two flashlight batteries for two years. And it's cheap. It's like a buck or two on bill of materials. So what you can do is replace mice and keyboard cables. You can connect your thermostat. It wouldn't be used for video or high-quality audio, but it is very inexpensive, and I think that will drive connectivity into a lot of areas where cost or battery time issues would have prevented it in the past. You will start to see samples on that late this year, early next year. You'll see products in 2003.
How about smart cards. Are those mostly still only overseas?
Smart cards are more popular in Europe than they are in the U.S., but a lot of things will drive the growth around the world. They have become very popular for employee badges. We see a lot of nations and a lot of applications where people care more about security. You can drive the price of this down to so you can embed it in paper tickets for airplanes, or luggage tags or even simple things like embedding these things in rental movies or library books. We have a project going on in Singapore, where you can check books out of the library and deposit them in any bin located around the city--and they are automatically checked in when you toss them in the bin!
We have a major retailer in the U.K. that is labeling all its materials with smart tags. We have a clothing manufacturer that we can't announce that is going to incorporate the tags in all of its products. The neat thing about it is that it is not only for inventory tracking and so forth, but they will also use it in the stores for checkout and tracking that. And it also guarantees the genuineness of the product.
There is also a shoe manufacturer looking at this because the knockoffs won't have the smart tags in them. We laugh at it now, but you could have a kiosk so when you walk by it says, 'Congratulations for buying those shoes.' There are different motivators. (RF tags) can allow you to inventory an entire shelf by scanning it. You can have an entire palette of products and push it through a gate and it will tell you everything in it. It can simplify checkout.
There's a lot of interest in PVRs, too (personal video recorders). In the future, chipsets will include TV tuners so that there effectively will be a PVR with every PC.
There is no reason you have to 100 percent own your own fab anymore.
We believe that PVR is something that would be logical to build into the TV itself. If you look at the TVs coming out, you basically get the PVR capability almost for free because it has the MPEG encode and decode in there anyway. So all you have to do for a PVR is either add a hard disk and/or add a recordable DVD device.
One of the things that will work in favor of the TV is that the PC industry has done a great job of driving common standards that drive down the cost of I/O. All disk drives cost $100 or less these days. (At that price) you can start thinking of your television as a device you accessorize rather than as this sort of fixed function box that you never upgrade or change.
Because of the nature of the embedded market, do you have to worry about a huge multiplicity of products, or are some components reusable from one product to another?
There is a very high leverage between all of these different things. You look at displays, when you look at video streaming. The PVRs we talked about--how TVs are going to move forward in the future. We reuse a great deal of our technology. It is a pretty small step from having an FM radio in a cell phone to having a TV in a cell phone, and in the past the displays in cell phones were so bad you wouldn't ever think of putting a TV in it. But when you look at the displays, going forward in cell phones there's no reason you wouldn't just want that as a checkoff functionality.
I imagine your ASP (average selling price) is pretty low because a lot of your chips go into products where you are not buying the other thing, not the electronics.
Well, yes and no. Take smart cards for example. The customer we sell them to gives them away for free to their customers, but it doesn't mean our ASP is low. We have prices that range from a small fraction of a cent to $2,500.
How do you handle your own manufacturing? Do you do a lot of it yourself or do you farm out to foundries?
We do about 80 to 90 percent of our manufacturing internally. However, going forward we've adopted a relatively new model; we have extended our relationship with ST (ST Microelectronics) for process development to include Motorola and TSMC (Taiwan Semiconductor Manufacturing Co.). The four companies are working cooperatively to develop 90 nanometer and beyond process technology. That opens up a very attractive possibility to create manufacturing facilities together or outsource to our partner TSMC.
Is it too difficult and too expensive to own your own fabs anymore?
I think the equation shifts a little bit. In the past, people viewed the process technology as a key differentiator, and it was part of the intellectual property and how they competed. You still do that in specialty processes, but in CMOS technology the success of the foundry organizations, such as TSMC and UMC, has basically made it where that is not an area where people compete anymore.
There is no reason you have to 100 percent own your own fab anymore.
Are there other foundries coming on line? Right now, there are basically only three major ones.
There are a number in China coming up, and a number of them are quite attractive. ASMC is one that we work with. We are the largest shareholder, and that allows us to get local content in China, which is valuable.
When will the recovery occur?
There are a lot of niches that are going to grow well, and the recovery is going to depend a bit on the overall consumer confidence and IT spending. I think the cell phone business is going to soften. I don't think we are going to see the 30 percent per year growth rates that we saw in the past. I think it will become more like 8 percent to 12 percent growth per year, longer term. I think people will go from replacing their phone every two to three years to every four or five years.
The new markets that cell phones will penetrate--we see as price-sensitive markets. Many of the emerging markets will be the targets there, because in Europe and North America we see saturation.
How about 3G? Will that drive demand?
3G has a tough rode to hoe here because there are a number of challenges. One is battery lifetime. The 3G battery lifetime isn't quite good. We will have to see how consumers react to that. Another thing is what will consumers do with all that bandwidth. It's cool to send movies to each other, but it's not that convincing.
There is also the emergence of things like the GPRS (general packet radio service) 802.11 convergence where you put 802.11 in your phone and you use that to connect to your laptop, or you just have 802.11 in your laptop and that takes away one of the big reasons people wanted 3G, which was for connectivity for computing applications.