Coffee Break with Gary Smith
I have known Gary Smith for almost 20 years: we met when he was a methodologist at LSI Logic and their salesman for 3Com dragged him in to encourage us to continue using the proprietary LSI tools and not move to Synopsys Design Compiler and Verilog for ASIC design. Like all good ENTJ‘s he was hearty, robust, and argumentative. I guess it takes one to know one. What Gary and I considered to be a friendly but spirited conversation the salesman was convinced had escalated to a shouting match. We have been friends ever since.
Much has been made of Dataquest’s decision to shut down their 5 person Design and Engineering Group, which covered EDA, electronic system level (ESL) design, embedded software, mechanical design, and architectural engineering and construction. I think Gabe Moretti‘s Life Without Dataquest (Oct 23, 2006) in CMP’s EDA DesignLine probably had the best insights (links added):
It is always sad to see people that have dedicated their skills to make a difference in our industry get laid off. It has happened too frequently in the last two years in the media, and the trend has now extended itself to the world of analysts. Some individual analysts, Erach Desai for example, have started their own consulting organizations instead of being associated with an investment banking firm. Now the news that Gartner Dataquest has decided to close its CAD group at the end of this month and terminate coverage of EDA highlights once more the real nature of the industry.
When three of the handful of publicly traded EDA companies control about 73% of the industry revenues as Merrill Lynch Research’s Jay Vleeschhouwer stated in his last report, one only needs to look at Cadence, Synopsys and Mentor, to determine the state and direction of the industry. By adding Magma to the input data one has enough to develop a sufficiently accurate picture of the industry to advise EDA corporate planners and independent investors on both growth opportunities and possible pitfalls. Gartner has thus made a difficult but financially sound decision in terminating its EDA coverage. In spite of the excellent professional qualifications of each of the members of the CAD group, and the technical leadership they provided under the able guidance of Gary Smith, the business side of this enterprise has become less justifiable. The four largest companies have grown an internal knowledge of the competition and thus have lessened their reliance on the input from Dataquest, while few of the smaller companies have developed the marketing bandwidth to take advantage of the private analysis offered by Dataquest.
Just as the amount spent on advertising and the changing nature of the ads have shaped the reorganization of the industry coverage by media giants like CMP and Reed Elsevier, so the amount of money spent on independent market analysis is reshaping this segment of the industry. I have no doubt that companies will continue to rely on independent input for their planning, but this help will come from individuals or small organizations that can operate with a significantly smaller overhead than Gartner. If you really think about it, supporting capabilities, like media, PR, and market analysis, are now resembling the nature of the industry they serve: innovation and creativity come mostly from small companies and individuals who find a way to believe in their own ideas and convictions and who are successful for what they know and contribute, not who they work for.
I like the Moore’s Law waits for no one sentiment he closes on. He was the first to really explain it as a rational business decision instead of some kind of conspiracy. John Cooley offered a wealth of them in the “Untold Gary Smith Back Story” but Steve DiBartolomeo of Artwork Conversion Software made what I think was the most prescient comment:
What’s replacing Gartner? DeepChip, and blogs. An EDA company makes an outrageous claim? Within hours actual users will refute such claims on blogs, emails to DeepChip and the bragger is called to account. A major customer changes suppliers, the news is out in days. A new tool is crap, in spite of the NDAs enough “Call Me Anonymous” engineers report their experience. The services that you provide freely via DeepChip compete directly with Gartner. Information zips around much more freely than before; yet someone has to aggregate it, qualify it, filter it and make sense of it, but it is pretty clear that the Gartner business model has reached the end of its life.”
- ANST – Ansoft – 5 analysts
- ARMHY – ARM Holdings – 4 analysts
- CDNS – Cadence – 11 analysts
- LAVA – Magma Design Automation – 7 analysts
- LVGN – Logic Vision – No analyst coverage
- MENT – Mentor Graphics – 7 analysts
- MIPS – MIPS Technologies – 4 analysts
- PDFS – PDF Solutions – 6 analysts
- SNPS – Synopsys – 11 analysts
- SYNP – Synplicity Inc. – 3 analysts
- VIRL – Virage Logic – 5 analysts
There is a fair amount of overlap in coverage but it looks like there are perhaps 16-18 analysts covering at least one company in the industry and a core of about a dozen covering at least three. As a contrast, Xilinx has 27 analysts covering it and Altera has 30. These two FPGA players probably invest as much in CAD tools as many if not most of the companies listed above.
Peggy Aycinena interviewed Gary in 2001, posting it in 2004 here where she and characterized him as the EDA Industry’s answer to the Oracle at Delphi. While he has survived a bout with cancer and has a young son still in diapers I have never witnessed him display any supernatural powers. He is strong methodologist who is able to spot the part of the future that’s already here by observing and listening to designers.
As to what Gartner’s decision means for the industry, I am guided by Gerald Weinberg’s observation that “It may look like a crisis, but it’s only the end of an illusion.”