IC Vendors Vary Widely in Their Projections of Digital Power Revenue

Perhaps it's fitting that when power IC vendors were recently asked about the near-term prospects for revenue from digital power technology, their market projections varied by at least a few ones and several zeros. Although they weren't answering in binary, the industry executives who participated in the digital power rap session, "Is there Money in Digital Power," at APEC last month cited widely varying sales figures when asked about the market for digital power products. Revenue projections ranged from less than 100 million in 2006 to as much as $10 billion in 2010.

The disparities expressed by panelists from Zilker Labs, Silicon Labs, Texas Instruments, and Intersil, reflected their very different perspectives on the digital power market as well as their different approaches to quantifying market potential.

Brett Etter of Silicon Labs put forth the most optimistic assessment of the market in citing data gathered by Darnell.^* Their data projected digital power revenue from isolated and nonisolated dc-dc converters as well as merchant ac-dc power supplies. Excluded from the projections were PC power supplies and voltage regulator products specifically developed to power microprocessors (i.e. the VRMs and VRDs). But even with these products omitted, Darnell projected that the market for digital power supplies in 2010 would be at least $8 billion according to their "moderate forecast." And that number rises to nearly $10 billion in 2010 under Darnell's aggressive forecast.

Looking at the near term, Etter indicated that revenue for this market would be $3 billion in 2008 and $1.6 billion in 2007. However, he qualified these expectations by explaining some of the assumptions behind these numbers.

"If you are a power supply manufacturer and you need to get to a 2008 number of $3 billion, you've got to start the design essentially today. It takes you 9 to 12 months to do that design, get it through qualification and ready for production. And let's say it overlaps a little bit with that OEM design in process - they take another 9 to 12 months to do their design in process. You're looking at revenue starting in '08. So that $3 billion has to start today," said Etter. And with respect to the $1.6 billion projected in 2007, Etter noted "OEM qualification needs to be starting pretty soon, if you're going to get in there and get that revenue started in '07."

But not everyone agreed with such high projections for digital power supply revenue. Dave Freeman of Texas Instruments said $8 billion in 2010 is "too high" and went on to explain why. "I don't think we have the infrastructure, I don't think we have the talent to start doing digital power supplies no matter how good we make the tools. The power supply industry is relatively slow moving. I don't see even a billion dollars in 2008. That would take divine intervention." When asked later about his comments, Freeman elaborated by saying, the industry lacks "the comprehensive talent needed to broadly use digital power."

When asked to project digital power revenue in 2007 and 2008, Freeman said, "Let's start with 2007, we've seen numbers of 300 million in 2007, we think that this is a little overzealous...it's probably less than half of that." In 2008, Freeman suggested the market for digital power might reach $300 million.

Jim Templeton of Zilker Labs also expressed a more modest assessment of the digital power market. "We looked at this problem and sliced it and diced it. Our conclusion internally was that somewhere between $900 million and $1.7 billion of opportunity exists for the solution space, the total available market that would benefit from....a digital solution." In this case, Templeton was referring specifically to revenue from silicon, noting that "overall power management IC market is on the order of 6 to 8 billion." So, according to Templeton, the prospect that the total available market for digital power will be $1 billion or more in 2006 is believable if you accept the numbers given for the power management IC market.

Templeton qualified his projection further saying, "Keep in mind, this is our assessment of the total application space that would benefit from a product like ours [the ZL2005]. So, it's highly biased toward what our product does. Our assessment is this is a 2006 number." When asked about the "served addressable market" - as opposed to the total addressable market (or possible market) for digital power revenue - in 2006, Templeton responded "I think it's pretty small by comparison....neglecting VRMs, my gut instinct tells me that it's somewhere in the $80 million range."

Templeton later noted that his company's views were much inline with those expressed by Dave Freeman of Texas Instruments. Templeton explained that if you take his projection of $80 million in market opportunity this year and extrapolate out to 2008, he sees the same types of figures ($150 million and $300 million) suggested by Freeman.

Ami Joseph of Thomas Weisel Partners, the moderator of the discussion, contrasted the projections coming from Texas Instruments and Silicon Labs and how these numbers represented different views on how digital power technology would be adopted.

"Now in 2008, we have a competing bid between 300 million and 3 billion," said Joseph. "That's really a very interesting slew of opinions with some people thinking the initial design wins will prove the concept and lead to customers moving to deploy digital power quickly across product lines versus a more gradual approach as certain problems are faced and certain problems are solved."

Given the panelists comments, it appears that they not only feel differently about how fast digital power technology will be adopted, they also define the market differently. Some focus on "total available market", while others address what they expect will be the "served market." Additionally, some focus more on power supply revenue, while others are addressing silicon revenue.

Even Etter, who presented the highest projections for digital power revenue in power supplies, noted that his projections were " the actual power supply numbers from Darnell, the silicon numbers... are going to be much lower."

Greg Miller of Intersil also indicated that his company's sales of digital power chips would be modest this year. When asked about the approximate range of sales for his company's digital power ICs, excluding the results of the partnership with Primarion, Miller indicated results were somewhere in the vicinity of the "less than $10 million threshold" this year.

In the course of their presentations, the panelists cited a variety of factors that are expected to drive adoption of digital power. The growing number of supply voltages was one factor. Requirements for supply sequencing, monitoring, fault prediction and detection were others. Some pointed to demands for nonlinear control, optimization of power supply efficiency, and dynamic reconfiguration of the supply. The need for greater integration was considered another incentive for adopting digital power control.

Many of the reasons cited for adopting digital power relate more to digital power management (digital communications, control, and monitoring of the POL) than digital power control (digital control of the PWM). But Zilker Labs' Templeton summarized why many believe digital power control is needed, "Our view is that the real impetus for the uptake of digital power isn't because it's digital, but because it's really combining and solving these power management functions. It just so happens that an elegant way to do this is to combine it [power management] with the power converter itself."

However, panelists noted that a number of barriers exist that will slow adoption of digital power control. Etter noted that "analog control loops are still mainstream," because they offer lower bill-of-material costs, best in class performance, and more user-friendly design tools.