Notebook weight is a big issue to Anand Chandrasekher, vice president and general manager of the Mobile Platforms Group at Intel.
Lighter, thinner notebooks are part of the reason laptop sales are growing far faster than desktop sales. That shift is why Intel's quarterly financial report reveals that around 70 percent of the company's revenue now comes from overseas. And as if Chandrasekher needed further incentive, he personally spends a lot of his professional life toting a computer bag as he hustles through sundry international airports.
Earlier this year, Intel launched one of its most ambitious projects in the notebook market to date. The Centrino chip bundle includes the Pentium M, a new processor that conserves energy better than earlier Intel notebook chips, and a wireless module, a new product category for the Santa Clara, Calif.-based chip company. Chandrasekher sat down with CNET News.com to discuss Wi-Fi, the spread of notebooks, and why U.S. consumers still buy some of the heaviest machines on the market.
Q: What's the rate of acceptance on Pentium M compared to Centrino? Are most manufacturers picking up both or just the Pentium M?
A: For Centrino in wireless notebooks it's quite high. It is well over 50 percent. You've got to keep in mind that there are some corporations that have no wireless. In those environments they don't want any wireless coming in.
One thing that has been interesting to watch is that some Intel executives have said that Wi-Fi chips are a commodity already. But you had to delay your own homegrown Wi-Fi chip a couple of times. Is it trickier than you thought?
There are two aspects to this. We're putting a lot more time into validation and qualification of the product than our competitors. And it's for a very simple reason: We are building a brand that has wireless associated with one of our processors. If there is an issue with any one component, the entire brand can get hit. So we basically stepped up our validation efforts on wireless an order of magnitude.
The second reason is that what is perceived as a delay really isn't a delay. We got slammed in the industry for not being there when 802.11g was announced and other people had products. Well, the fact of the matter is that the "g" standard wasn't ratified until summer.
We learn lessons the hard way, but we learn them. You remember the RDRAM (Rambus DRAM) situation of early 2000? We put it out and the standard wasn't quite nailed and then we spent a lot of time debugging the spec in the marketplace, which is not a lot of fun if you are trying to ship millions of units at the same time. And we swore that at the time that we will never do that again. It's the same thing here with (802.11g). We got a lot of push to go forward with the product and a couple of our competitors started shipping but we said no, it is not a standard yet.
I am very comfortable with our 802.11b/g product. We will be in production in (the fourth quarter) and we will launch it in the (first-quarter) time frame.
What is involved in the validation process? What do you do, test your chips against other access points?
I'm past 40 now and have got to start worrying about back problems.
It is a whole range of things. You have to test the minicard, obviously, against the parameters it is designed to: voltage, thermal--all that sort of stuff. But then you also have to test it in the system. How does it do when you throttle power? But that's not sufficient either. We then have to test it with access points. Where's the access point located? In a room, the behavior is different--so you try to characterize all that.
And it is not just the access point. There is the backhaul. In the backhaul, if you have a server that is taking all of the traffic and it is a Linux server versus a Microsoft Exchange-based server--those two scenarios actually give you different results. You have to test against all of these permutations and then try to determine how much of this was caused by hardware or whether the hardware was irrelevant.
Roaming is a big issue for wireless. Users want to hop onto different networks. But carriers haven't really implemented it much. Is it a technology problem or a business issue?
It is a business negotiation issue. If I have a large network and my customers want to roam on your network, it basically means I am borrowing something you could have used. The question is how to evaluate the opportunity cost of that capacity. If I have a larger network, my tendency might be to say that my opportunity cost is larger, therefore you should pay me more.
The good news is that those issues are being resolved. T-Mobile and Wayport and STSN have agreed they are going to roam. They don't really compete. STSN is in hotels, Wayport is in airports, and T-Mobile has public hot spots, so they can cooperate, but I would like to see more of this.
I just spent time in Japan and Asia-Pacific. Out of those 10 days, I spent four days talking to carriers--not notebook manufacturers, but carriers. The interest is absolutely there. There is an organization that has been formed, the WBA (Wireless Broadband Alliance), whose focus is to facilitate roaming. These guys are going to start to do roaming internationally initially.
Oh yeah, you have Telefonica in Spain and...
Exactly. You aren't taking anyone's business away. I think it will start there and build a basis of trust from which you can get the domestic roaming issues worked out. It is a fairly new organization. We had a parallel organization in Singapore that is trying to get the technology issues resolved with respect to roaming.
When do you think we are going to start seeing notebooks that can switch from one band to another, like Wi-Fi to cellular, fairly seamlessly?
It will vary by geography. Asian carriers are all over this. In Japan and Europe, same thing. I don't see the same situation in the U.S.
Why is that? Is it the history of cell phones here, or is it that people are on the road more in other geographies?
I wish I had a pithy answer for you, but I don't. The interest level, the motivation to resolve these technical issues--I see it more overseas than I do here. I think you will see a solution in 2004 and a lot more solutions in 2005. By the time we get to 2005 this issue will be a no-brainer.
Next year, with Centrino and Banias Celeron, we think retail notebooks will change.
Mostly you will see it as a build-to-order solution. When you go to wireless WANs, you have to be certified and go through regulatory procedures. You already have it with wireless LANs, but when you go to wireless WANs it's an order of magnitude more complex. The network operations want to make sure your device isn't going to bring down the entire network. The amount of time to certify something is several months. But at the rate the notebook industry moves, if you have to wait several months, you have just missed several product cycles.
You have a bunch of stuff coming into notebooks in the second half of 2004. What is that going to do to notebook design?
They will get thinner, but it's also weight. The PCI Express card is 50 percent smaller. And if everything around it is the same way, it will weigh less. I'd pay for less weight. I carry around these things all over the world. I'm past 40 now and have got to start worrying about back problems.
Check out a new Panasonic notebook. It's about 1 inch thick, and you think it should be heavier, but it is very light. The way they achieve that is that the chassis has a bunch of ripples in it. You think they are there for cosmetic reasons, but it's for mechanical and structural integrity. In fact, in strategic places they give it holes. It's counterintuitive. Why would you punch a hole in it? But it gives it flexibility.
How about the extra-screen concept, where you can shut down the notebook but still use a small screen to send and receive messages?
You'll see a few in 2004 that do that. We already have a few OEMs (original equipment manufacturers) that are going to design that in. 2005 is the time when you will see a lot of it.
Now onto Dothan, the next Pentium M. When does it ship?
We will be in production in (the fourth quarter) and will ship to OEMs. We are bringing out the Celeron version of Banias. (Banias is the code name for the processor core inside the Pentium M). That will be sometime in the (first-quarter) time frame.
Next year, with Centrino and Banias Celeron, we think retail notebooks will change. You will see these thinner, lighter models. Interestingly, if you go to Asia, Europe or Japan, you already see the thin-and-lights in retail. You go to the United States, you see the fat (models). There's not that much of a cost factor with thin-and-light. If you are trying to do a 1-inch (thick notebook) there is, but not if you are trying to do a 1.3-inch. Between a 1.8-inch and a 1.3-inch, the cost delta is fairly small.
If you guys get a chance to go to Japan, walk through Akihabara. You think "why the heck aren't these in U.S. retail?" That logjam will get broken, and I think '04 is the time it will get broken.