The faster processors of increasingly powerful computers are starting to generate enough heat to melt circuit boards -- a problem that's more and more difficult to manage. But as often happens, one company's problem is another's opportunity. In this instance, International Rectifier is the beneficiary. The El Segundo (Calif.) outfit makes chips that manage power consumption and heat in electronics products. The higher the heat, the hotter IR's sales.
On the basis of revenues, January was the best month in IR's (IRF) history, says Alex Lidow, its CEO. That's saying a lot, considering that IR reported 20% year-over-year growth for its fiscal second quarter, which ended in December. As its sales soared to $252.3 million in the period, it swung from a year-earlier loss to a net profit of $16.8 million, or 26 cents a share.
Results could improve even more as IR ramps up sales of its power-management technologies for plasma TVs, washing machines, laptops, and cars. That's why three brokerages have upgraded the stock to a buy since January. It trading about 22% below the 52-week high of $56.90 it hit last fall. Analysts polled by Thomson One expect IR to report revenues of $1.03 billion for the year that will end in June, up 19%, and net income of $99 million, or $1.51 a share, vs. a loss of $90 million the previous year.
How does IR solve heat problems for customers such as Intel (INTC), for which it makes power-management chips? Lidow talked about these and other issues with BusinessWeek Online reporter Olga Kharif on Mar. 4. Here are edited excerpts of their conversation:
Q: We've been hearing a lot about processor heat problems lately. What's going on?
A: Typically, chips double their transistor density every 18 to 24 months. Those improvements have been driven by a shrinking of photolithographic line widths -- essentially, the width of a ray of light used to make chips. If you cut the line width in half, you can quadruple the number of transistors in that space -- and your chips' performance keeps going up.
But it was clear back with Intel's Pentium II [in 1997] that a thermal issue was threatening to stop [gains in transistor density]. The more transistors you have on a chip, the more electrons it takes to feed them. And the faster a processor's clock speed -- the faster it can crunch numbers -- the more electrons it takes. The electrons dissipate their energy and produce heat.
Basically, the amount of heat generated on a chip is doubling every 18 to 24 months. A while back, Intel did a graph that showed the Pentium 4 was running at the temperature of a hot plate. If you extrapolate, by 2007 Intel's Pentium generation would be at the temperature of the surface of the sun.
Q: So what can be done to fix that?
A: The heat is generated by the number of electrons flowing into the chip each second that then lose all their energy. If you put electrons into that chip [in such a way that each has] less energy to lose, the chip ends up generating less heat. So Intel's roadmap has shifted from shrinking the line width to lowering the voltage -- the amount of energy carried by the electrons. Ten years ago, Intel microprocessors operated at 5 volts. Now, they're nearing 1.3 volts. And by 2007, they'll go as low as 0.6 volts.
Q: How will that be achieved without hurting performance?
A: Intel has been leading the industry with the use of new chipmaking materials that allow for [lower voltage]. Processors made from those materials need to be combined with our power-management chips to allow for superior performance of the computing system.
It's analogous to living in a house that gets water from a water main. Suppose that every 18 to 24 months, twice as many people want to take showers at the same time -- even as somebody at the main line is turning down the water pressure. The only way you can deal with this is put in bigger pipes and booster pumps -- and that's the function our chips perform.
By using special transistors and integrated circuits to deliver electricity, they essentially make sure that key chips on the motherboard get just the right amount of power for maximum performance and generate less heat -- and that the system as a whole doesn't overheat.
Q: What are you seeing in terms of demand for your chips?
A: Power-management content went up dramatically between the Pentium 1 and Pentium 4. It's not just the processor but also the graphics chip and memory that increase the need for our products. Now, there's a gigabyte of RAM in low-end PCs -- and that takes a lot of power. Based on an average growth in PC unit sales of 10% a year -- the average for the past 15 years has been 15% -- the demand for power management should rise about 22% each year.
Of course, there's no such thing as an average year. Last year, PC unit sales grew 14%, and our shipments into PCs grew 73%. Some of that was due to our ability to gain market share. But a lot was due to increased demand for power management in notebooks that use Intel's Centrino chips.
Q: In portable devices, such as laptops, heat is a much bigger issue than in desktops. How big is your opportunity there?
A: In 2003, our sales of chips that go into laptops grew 86%, while laptop unit shipments grew 26%. People who use laptops demand the same performance as they get on desktops. They also demand 8- to 10-hour battery life.
Q: Why are you not that interested in selling your chips into other portable markets, such as cell phones and personal digital assistants (PDAs)?
A: Today, that market is less than 1% of our revenues. The reason is: A cell phone or a PDA needs less power management.
Here's why: People buy computers because of microprocessor performance. Therefore, the microprocessor is pumped to the max -- and what determines that maximum is how hot you can let that chip get. So, the entire design is thermally limited.
But you aren't using cell phones and PDAs to do heavy calculations and spreadsheets -- yet. People have low expectations for the graphics on these devices, so they aren't thermally limited. As they become more like laptops, they'll begin to require more power management. We're just not there yet.
Q: When do you think we will be?
A: I think 3G [third-generation] or 4G wireless networks will be deployed before you get close to a thermal problem in phones and PDAs. You need high-bandwidth wireless connections that such networks will offer before you need more powerful processors. And 4G deployments are slated for 2010.
Q: Many consumers are concerned about energy efficiency. Are they willing to pay for it?
A: We have pioneered technologies that can make motors in refrigerators, air conditioners, and washing machines save half the energy they use today by varying the speed of the motor -- for instance, when you open the refrigerator's door. That's great, but most people don't choose an appliance because of its energy savings but because of its functions.
So we designed several new technologies for appliances that are both energy-efficient and cheap. Washing machines based on our technology are ultra-silent, they wash clothes better, more gently, and in about half the time. And these washing machines are hot sellers.
Sales of refrigerators, air conditioners, and washing machines are 200 million units a year and are growing slowly. But the conversion rate to silent, fancy versions of all these is about 19% per year.
Q: Some people buy cars for their energy savings. What are the opportunities for your company there?
A: We can help save half the fuel burned by cars -- at no cost to the consumer. You have to implement just a few things, including the use of a hybrid drive, which allows the engine to stop whenever the car stops, so you don't consume energy during idling. When you get going again, an electric starter motor allows you to accelerate, and then the engine starts up and takes over. We can basically give a consumer a price-neutral, performance-neutral, double-gas-mileage vehicle.
Here's where we come in: Many of the motor drives that go into the cars are exactly the same as what we've pioneered for refrigerators and air conditioners. This is a $10 billion potential market for us.
Q: How dependent are you on the general semiconductor cycle?
A: We're somewhat buffered from the semiconductor cycle. Demand for a lot of our products, such as drives that go into washing machines, is growing faster than the market for those devices because of conversions to ultra-silent models as well as because various devices are requiring more of our chips. Let's say sales of washing machines decline 3% vs. growing 3%. The growth rate of the corresponding part of our business will fall from 19% to 13%.
Similarly, in bad times people buy fewer cars -- but they also buy more economical cars [such as hybrid vehicles]. And in IT, high-end and portable PCs have been less volatile than the rest of the market.
Q: There's been a lot of speculation about how long this semiconductor upturn will last. What's your take?
A: We had significant delayed capital spending on new semiconductor factories and equipment for the past two years. Once everybody starts investing in new factories, you normally have an 18- to 24-month lag before there's overcapacity, and the market collapses. So we'll see probably 18 to 24 more months of an upturn.