My previous post didn't make it so easy to access the article I was alluding to. Here's the whole thing. My primary fear for Gennum in this is not so much that they will particularly mismanage Nano, but rather that the key Nano people, being entrepreneurial types, simply will not be inspired working as employees of a little chip company based 7,000 km away. They'll know that Burlington is, and always will be, the centre of the Gennum world. Highly sensitive IP and other info will flow from Bristol to Burlington but only selectively from Burlington to Bristol. I like the Gennum people and the existence of the company is a credit to Canada and so I hope this move doesn't wreck it. But it might.
========================
What went wrong with MIPS-Chipidea merger?
Posted:20 May 2009
Most mergers and acquisitions rarely work. When they do, it's because both parties have time, patience, commitment and money to burn in the process of integrating the two organizations. An examination of the recent MIPS Technologies-Chipidea breakup strongly suggests that neither one had sufficient amounts of these key attributes.
Only 18 months after the Chipidea acquisition, MIPS—faced with quarterly scrutiny by investors and the sharp decline of the global market—had to give up the
analog and mixed-signal IP supplier, once regarded as Europe's best kept secret.
MIPS executives blame the failed marriage on the overall softening of the economy. John Bourgoin, MIPS president and CEO, nonetheless stood by his original view, calling the acquisition of the analog IP vendor a "strategic move."
Inside scoop
In contrast, former Chipidea engineers and marketing officials said "MIPS and Chipidea was a good fit, but [the
merger was] mismanaged." They argued that it was MIPS' "integration dogma" that destroyed Chipidea's sales organization, thereby undermining the merger.
After the recent posting of an
EE Times' article on a LinkedIn networking group for former Chipidea employees, several agreed to talk to
EE Times on the condition of anonymity.
Integration was a key issue from the start, they said. MIPS licenses its digital processing cores. Chipidea licenses analog and mixed-signal IP. Said one former Chipidea engineer: "There was no overlap whatsoever in our businesses."
He offered the following analogy: "If you are an aircraft manufacturer, you design a plane, manufacture it and you're used to the lifestyle of selling one or two aircraft a year to guys with money." Then, "you acquire another company . . . but in order to make revenue, you suddenly realize that you'd have to sell 10,000 airplanes a year."
Following the acquisition, MIPS executives still believed they were in the same
semiconductor market segment, the former Chipidea employee said. "They never understood that the daily demand and daily grind of meeting so many different customers' needs in the analog business required another sales team."
Contrary to the claims of MIPS executives that "Chipidea, even after the acquisition, was operating as a standalone business with its own marketing, finance,
R&D and engineering organization," another former Chipidea employee said MIPS early on insisted on the integration of the companies' sales forces, "which turn[ed] out to be a disaster."
As early as Q4 07, MIPS executives already noted in conference calls with financial analysts "that the integration of the two sales organizations was not working," the source added. "Quarter after quarter, they kept talking about how difficult it was, but they absolutely did nothing about it."
It's not that Chipidea executives didn't understand the importance of integration, but in their view the best solution was first "to preserve what you have," then to assimilate.
Former Chipidea employees expected to retain a semblance of their former corporate identity while working to expand the merged company's business. They also hoped to become an integral part of "a virtual system semiconductor company," operating on the assumption that more analog IP blocks would be required in future SoC.
MIPS, recognizing the growing importance of analog technology to the semiconductor market, expected Chipidea to quickly contribute to its revenues. Indeed, many U.S. companies understand that acquisitions are a tactic for boosting next quarter's financial results.
Lisbon-based Chipidea generated more than $25 million in revenues in 2006, including an annual compound growth rate of 50 percent. The analog and mixed-signal IP market grew 34 percent in 2006, accounting for 16 percent of the overall design IP market, according to Christian Heidarson, a senior research analyst at Gartner.
In 2007, MIPS wasn't alone in pursuing Chipidea. Synopsys and others were also in the running to acquire Chipidea. The result was a bidding war, according to industry sources. Even though MIPS wasn't the highest bidder, Chipidea's CEO at the time reportedly pushed for the MIPS deal as a way to keep the company intact. MIPS paid $147 million in cash.
Earlier this month, MIPS sold Chipidea to Synopsys in an all-cash transaction totaling $22 million.
Bad move
Some industry watchers are critical of MIPS for buying Chipidea.
"Chipidea turned out to be a bad idea," said Linley Gwennap, president and principal analyst at The Linley Group. "MIPS picked the wrong company to buy and quickly discovered that what they thought was a profitable business was actually a sinkhole.
"Worse, dealing with these problems caused MIPS to take its eye off the ball in the CPU segment." According to Linley, "For MIPS, the sale of its analog business is a classic case of addition by subtraction—now the slimmer company can focus on its core business."
Technically speaking, "Although an SoC is likely to use both
processor and analog IP, the two types of functions are distinct," explained Joseph Byrne, another Linley Group analyst. "There's little technical synergy between them."
In other words, "any synergy derives from having a larger portfolio of products to offer each customer and from cross-selling one product into the other product's installed base of customers. A challenge for MIPS's management was to keep advancing MIPS processor technology while devoting time and attention to the newly acquired analog business," said Byrne.
"This challenge was compounded by the fact that the analog business faltered immediately after the acquisition and before the global economic crisis. For the December 2007 quarter, the first full quarter after the acquisition, the analog business group generated nearly $10 million in revenue. Revenue declined steadily in each of the five subsequent quarters, and in the March 2009 quarter it was only $5 million."
The merger had other potential problems that were apparent from the beginning. First, MIPS emptied its bank account to fund the $147 million transaction, then borrowed heavily from a local bank to fund continuing operations. MIPS, which had a healthy cash reserve until the Chipidea acquisition, ended up paying hefty premiums on its term loan, putting pressure on the company to cut costs severely. Cost cutting and the resulting impact on the company were exacerbated by recent market weakness.
Integration was bound to be a problem, as MIPS itself acknowledged in a Securities and Exchange Commission (SEC) filing. MIPS was extending its reach not just into the analog IP business, an area where it had limited knowledge and sales and marketing expertise, but was also assuming control of a company based in Portugal with operations in various other European countries, a region where the parent company had not previously operated.
"This is a substantially larger acquisition than any that we have previously completed and involves technology and products that are largely new to us," MIPS said in its SEC filing. The "challenge will be further complicated by the geographical distance between our headquarters."
That was not all. MIPS needed all the help it could get understanding and integrating its new business unit. But Jose Franca, founder and CEO of Chipidea, who could have offered a bridge between the old MIPS and the enlarged company, resigned before the transaction closed. Franca's departure deprived the combined company of his input on integration and marketing challenges.
Franca and other key personnel at Chipidea "could be important to our ability to advance the Chipidea technology and to effectively market and sell its products," MIPS said.
Former Chipidea executives took issue with a decision by MIPS management in July, 2008 to have Franca, president and general manager of the analog business group report to a new chief operating officer John Derrick, who headed the processor side of MIPS' business. By August, Franca was gone.
Franca was not available to comment, but former Chipidea employees saw the move as evidence of Chipidea's mistreatment by MIPS. They also pointed out that Derrick left MIPS in January.
MIPS argued that the analog market in general was declining by Q4 07, immediately after it acquired Chipidea. Some analysts dispute that assertion.
Dublin-based Silicon & Software Systems (S3), which acquired Acacia Semiconductor S.A., a data converter IP developer in October 2007, had a much different view of the market. Acacia was a much smaller analog IP company than Chipidea, and S3 was already in the analog IP business. It licensed its own IP to fabless chip companies. Similar to Chipidea, S3 saw its own analog IP business growing at a compound annual rate of as much as 70 percent between 2004 and 2006.
Dermot Barry, S3's VP of consumer silicon, said its analog IP business remained strong through mid-2008. "In the second half of 2008, the growth stalled," said Barry, "but I [saw] it stabilized."
S3 is expecting its design services business to decline in 2009. Barry, however, reported signs of a pick-up in analog IP negotiations over the last several weeks.
Barry speculated that fundamental differences in the business models used in licensing processor cores and analog IP was a reason for the failure of MIPS and Chipidea. "In the analog IP business, you need to have a much more specialized sales team, backed up by an engineering support team, which can quickly respond to any slight modifications their customers would require," said Barry. "In contrast, the processor business is, more or less a one-size-fits-all model."
Indeed, there is some evidence to back this assertion. Although some sources said MIPS had indicated that it would retain Chipidea's sales team, it ultimately decided to against the idea. Prior to closing the transaction in August 2007, MIPS had about 55 employees on its sales and marketing team. The group grew to only 59 employees at the end of the company's fiscal year on June 2008.
The engineering group also suffered as Chipidea trimmed payroll, even cutting some analog engineers it had touted as key benefits of the merger. Barry of S3 added, "We did profit from it by picking up some very good engineers from Chipidea."
Several former Chipidea employees said the market outlook for analog IP was healthy until the second half of 2008. It was the MIPS processor core business that got into trouble during that period, they said.
That claim could not be independently confirmed. However, it was clear that the MIPS-Chipidea transaction had a very short honeymoon. MIPS closed the transaction with limited cash on hand and immediately found itself with even higher costs related to a new analog IP division. While revenue soared in the year following the merger's closing, costs also climbed sharply higher.
MIPS closed its fiscal year ended June 30, 2008, with revenue of $104.8 million, up 26 percent, from $83.3 million in the previous fiscal year on the lift provided by Chipidea. License and contract revenue alone increased 50 percent to $58.4 million from $38.9 million. However, the cost of license and contract revenue shot up an astounding 55 percent of sales compared with slightly more than 4 percent in the preceding year.
This, combined with a $103 million writedown in the value of goodwill related to the Chipidea acquisition, saddled MIPS with a huge net loss of $131.8 million for the 2008 fiscal year versus net profit of $8.5 million in fiscal 2007.
Different path
Had MIPS shown stronger financial results in its core business, the marriage with Chipidea might have followed a much different path.
As analyst Byrne noted, in terms of combined licensing and royalty revenue, MIPS was established as the No. 2 supplier of 32- and 64bit CPUs, with a market share of 15 percent, well behind ARM's 67 percent share.
In terms of units sold, it was a much different story. Bryne said, "MIPS is still number two, but the company's 9 percent share is even further behind ARM's 76 percent share, and is close to ARC's 8 percent share."
While the discrepancy between revenue and unit share illustrates how MIPS' average selling price is better than its competitors, a softening of MIPS' core business remains a concern. "MIPS did not keep up with the growth of the overall market in 2008," said Bryne. Industry revenue and units grew 7.5 percent and 36 percent, respectively, according to a Linley Group report. MIPS's revenue was even y-on-y. "Core shipments were up 21 percent. While impressive, it still lagged that of the industry," Byrne added.
There is no question that MIPS, while highly profitable, "is threatened by ARM, which has a strong product line and dominant market share, and by smaller companies targeting niche applications or offering low-cost alternatives," said Bryne.
Upgraded processors will help MIPS serve its customer base and keep them from defecting to other architectures, he added. But "the company must also target new customers and applications to expand its business. MIPS must also work to regain mindshare for its architecture."
Based on Acacia's success with S3 and Chipidea's absorption by Synopsys, it's tempting to conclude that the analog IP business is, after all, a design service. When asked in 2007 if Chipidea was better off joining a design service company, then-Chipidea CEO Franca said, "We don't do design for hire."
Meanwhile, Gartner predicts that up to 90 percent of cores inside SoCs will be analog by 2011. Hence, the number of IP blocks to be supported will likely increase. Further, process technology spread among component and SoC products is increasing as a function that needs to be supported at a growing number of process nodes.
While many of these trends may boost external analog IP suppliers' businesses, the fact is that scaling the analog and mixed-signal IP business hasn't gotten any easier—before or since the MIPS-Chipidea fiasco.
-
Bolaji Ojo, Junko Yoshida
EE Times