Evolution of the board business - Embedded.com

Evolution of the board business


Over the past twenty years or so, we've seen an incredible amount of change in the traditional merchant board markets. But, none more dramatic than the Mergers and Acquisitions (M&A) activity involving board makers over this period. My records show that 76 board companies were bought and sold through the Spring of this year, starting with Intel selling their Multibus board group to Radisys in 1987. I'll try to walk you through the most significant ones here.

The sell-off began in earnest in the late 1990s, with five sales in 1998, seven in 1999, six in 2000, and eight in 2008. The peak year was 2004, when nine companies were bought/sold.

Looking closely at these transactions over the years shows some distinct patterns. Also, we can see that some large companies acquired board companies, only to divest them after a few years. Intel is a good example of this. It sold its Multibus group to Radisys (1987) and got out of the board business. Then, it bought Dialogic in 1999 (for about 2.6 times sales) and Ziatech in 2000 for about 3.4 times sales.

Intel then sold the Ziatech group to Performance Technologies in 2002, and sold the Dialogic telecom board group to Eicon in 2006 (along with Intel's Media Server Group). Finally, Intel exited the telecom board business by selling the remnants of its board group to Radisys (in 2007, for 0.50 times sales, plus cash for its existing inventory). The devastation in the telecom board segment is glaringly clear when you look at the prices these companies were purchased for, compared to what they garnered when sold off.

Other patterns of note include the entry and sustained activity of SBS, Curtiss-Wright, Mercury Computer, and GE/Fanuc in the increasing M&A activity. SBS bought Greenspring (in 1995), Bit 3 (1996), Logical Design (1996), OR Computer (1998), VI Computer (from Themis, 1997), SDL (2000), Essential (2002), and Avida (2003). It then sold the entire company to GE/Fanuc in 2006 for about 1.4 times sales. SBS initiated the M&A frenzy in the board business, but they rapidly became over-invested in telecom, and finally had to sell out.

Now let's look at Motorola's board group and its acquisitions. Motorola bought Prolog (1999) in their own preparation for the coming telecom wave. They bought Bluewave (2001) in a stock-swap deal) to get into the DSP-driven telecom segment. (Bluewave was a merger between Mizar and Loughborough Sound in 1997). Moto bought Paceline and Netplane (both in 2003) to get into serial fabrics on telecom boards. Then, they bought Force Computers from Solectron in 2004 to gain market share in the European ATCA market.

Three weeks after Intel exited the telecom board business and sold out to Radisys, Motorola announced the sale of their entire board group to Emerson (in 2007 for 0.67 times sales). Again, over-investment in telecom and the abandonment of traditional products and markets were the bricks that paved the path to divestiture.

During this rampant acquisition period, and during the great telecom love-fest, the Contract Electronic Manufacturers (CEMs) also got into the M&A game, with poor results. Solectron bought Force Computers (1996) only to sell them to Motorola's board group (2005). Celistica bought the HP's embedded group in 1998, and they were never really heard from again. Solectron also bought Smart Modular Computers in 1999. CMAC bought MIL/COTS board supplier DY4 Computers in 2000, and then CMAC got bought by Solectron in 2001. DY4 was later sold by Solectron to Curtiss-Wright in 2004.

In 2002, Solectron bought Compaq Computer's embedded group. Five years later, Solectron was sold to Flextronics for less than 0.30 times sales, according to the stock-swap prices at the time.

Curtiss-Wright, at the time, was focused on the MIL/COTS markets. Hence, it purchased Lau Defense in 2001, and the recently acquired VISTA. DY4 bought Ixthos (1998), then DY4 then was acquired by CMAC in 2000, and CMAC by Solectron in 2001.

Mercury Computer, a fairly diversified company, got into the M&A game in 2002, when they bought Myriad Logic. In 2004, they bought Momentum Computer and TDS Group. In 2005, they bought Echotek. The focus for Mercury obviously centered on MIL/COTS suppliers.

GE/Fanuc is also a diversified group. It bought Computer Dynamics (1999) and VMIC (in 2001) to expand in the industrial controls marketplace. In 2005, it bought the remnants of DNA Computing, and in 2006, it purchased Condor Engineering to strengthen itself in the MIL/COTS markets. It bought SBS (2006) to expand into telecom segments. Later that year, it acquired Radstone, who had previously purchased ICS and Octek.

In the industrial market Kontron bought Teknor (1999), PEP Modular Computers (2000), Micro Automation (2001), Jumptec (2002), and a division of Dolch (2005). Its rugged mobile business was sold to Crane (2007) as Kontron bought Thales Computer (early 2008). The Thales acquisition got Kontron into the MIL/ COTS segment, providing some diversification.

Emerson, a new entrant in the board markets, bought Artesyn (2006). In late 2007, Emerson bought the telecom-focused Motorola board group (for about 0.67 times sales).

Packaging/backplane industry wants in
Other M&A activity was going on in the packaging/backplane industry. In 1998, APW bought Vero Electronics, and Carlo Gavazzi bought Aurora. In 2001, Rittal bought Kaparel, Elma bought Trenew, and Carlo Gavazzi bought Channel Access. Elma bought Optima in 2004, and Mektron in 2005. That same year, Schroff bought the Electronic Solutions packaging group from APW.

If we look at the trend-lines of multiples over time, we begin to see some interesting directions. Telecom board companies sold for as much as 3.4 times sales back in 2000. But, the sale of Intel's and Motorola's telecom board groups in 2007 (at 0.5 and 0.67 times sales, respectively) show a distinct devaluation of companies in this market segment. On the other hand, MIL/COTS board companies, primarily focusing on VME technology in military programs, were selling for about 1.4-1.6 times sales in the early 2000s. But that's risen to over 3.0 times sales in the last few years. In the industrial segment, companies were selling for 1.0-1.3 times sales in the early 2000s, and the latest transactions are showing evaluations of about 0.7 times sales.

These trends demand explanation, and I believe I have the answer. As a rule-of-thumb estimate, a company is worth about five times its gross profit margin (GPM). It's clear that the GPM of telecom board companies is low. Competitive forces are at work here, but these telecom-focused companies got into the “systems” business, and the GPM on systems is inherently low in any of the computing segments. Systems builders make their money on volume shipments, but those volume shipments never materialized for the board vendors in the telecom space. Hence, their margins are low, in the 10% to 15% range.

The MIL/COTS VME suppliers typically show margins from 50% to as high as 70%, which is why we see MIL/COTS VME board suppliers now selling for more than three times sales. In the industrial segments, those markets have been transitioning to PC-based technologies since the 1990s. Again, there are competitive factors at work, but PC-based products always have low margins, somewhere in the neighborhood of 10% to 15%. Hence, telecom and industrial market segment margins are low because of commoditization and highly undifferentiated products. This is why lower multiples were paid for those companies–because of their low GPMs. In the MIL/COTS segment, the products have high margins and they're differentiated. Hence, the higher multiples paid for these companies. Once a specific vendor is designed-into a military program, that order is solid and generally not subject to future rebidding.

In addition, the telecom board segment growth and size is limited by the CEMs. Once the volumes come into their interest range, CEMs will build telecom boards for 7% to 8% GPM. None of the established board vendors in this segment can compete with that, and none of them offer any real value-added, compared to a CEM.

Ultimately, no board-level, hardware-oriented company can ever sell for more than 3.75 times sales. Why? Because board vendors have low fixed costs, but they can't get their variable cost (their direct cost of components) down below about 25% of sales. That gives them a maximum GPM of 75%. Five times 75% is 3.75 times sales, the maximum that the model says someone would pay for a board company.

On the other hand, software companies sell for up to nine times sales, as we saw when QNX sold to Harmon Automotive back in 2005. Why such a high multiple of sales? Because, software companies have variable costs near zero (i.e., they have no component costs in the selling price of the product). But, they have high fixed costs. Software companies make their money by getting low-cost, long-term financing. Their variable costs are already at or near zero, making their GPM very high.

In the next few years, I expect to see the purchases of VME-based MIL/ COTS companies continue. These are the best companies to buy from a GPM standpoint, with guaranteed cash flows and growth opportunities. I expect to see telecom board companies sell at fire-sale prices, as we saw in 2007. I expect to see dramatically increased M&A activity in the industrial board segment, and I also expect to see them sell for low multiples.

Ray Alderman is the executive director of VITA. He can be reached at .

1 thought on “Evolution of the board business

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.