LOS ANGELES Poor collective first quarter sales by the main EDA vendors have caused market research firm Dataquest Inc. (San Jose, Calif.) to slash its growth forecast in half for the 2000 EDA software revenue market from 15 percent to 7 percent. This would result in the software tools revenue part of the industry, separated from design services or hardware sales, rising to $2.765 billion in 2000.
The correction came as the 37th Design Automation Conference opened with workshops and technical meetings on Sunday June 4 and as Dataquest gave its annual state-of-the-industry address to assembled executives of the EDA industry.
"1999 was not a good year for the introduction of tools. At 3.5 per cent I believe it is the lowest growth in EDA [software revenue] since Dataquest began tracking the industry," said Gary Smith, chief EDA analyst at Dataquest. "Well, the top four EDA companies, which account for about 60 per cent of the market, sustained a negative 15.8 per cent growth in the first quarter of 2000. On that basis we ain't going to make our numbers in 2000, and I've cut the forecast from 15 percent to 7 percent. The only company that showed significant growth was Avanti," said Smith.
In an interview after his presentation, Smith said part of the problem was a collapse of sales at Cadence Design Services, due to the departure of sales people.
"Ray Bingham [chief executive officer] of Cadence is doing good stuff on the administrative and R&D side, but a lot of the sales people were Joe's," Smith said referring to Joseph Costello, former president and chief executive officer of Cadence, who left the company several years ago.
He said the EDA industry's problem was rooted in a failure to deliver on marketing promises because of bottlenecks in engineering. "The need is there, the demand is there, we've got to get some tools out there. The trouble is we've been losing engineers to these dot-commies," he argued.
Smith said that EDA companies, struggling with engineering shortages and a lengthening R&D cycle, were trying to finesse customers by showing them "the vision" and then delivering tools that fulfilled only a portion of that vision. Smith said that in the best cases it meant that money was left on the table but in the worst case it left customers feeling abused and with the impression that they must return to developing their own EDA tools.
In his presentation Smith observed that one reason for such problems at the electronic system-level (ESL) was simply that such tools were difficult to build and deliver. As a result, he predicted that RTL would remain the dominant form of EDA tool and by 2003 would have an annual market worth five times either the gate level or ESL tool markets predicted to be at about $275 million.
But Smith also showed a curve based on the supposition that ESL tools worked and experienced similar growth to RTL tools in their early years. The "what-if" r esult showed a 2003 market size for ESL tools of $811 million in 2003 or three times the current prediction.
Smith's was the last of four Dataquest analyst presentations. The meeting opened with Bryan Lewis, principal analyst and director of the ASIC/SLI program at Dataquest.
Lewis' theme finding was the winners in the ASIC market for 2010. In terms of market growth, Lewis was much more upbeat than Smith but he pointed out that the system-level integration (SLI) trend, sometimes called system-on-chip, was starting to have a consolidating effect on the semiconductor suppliers. For all but the first tier semiconductor integrated device manufacturers and semiconductor foundry operations, Lewis' message was stark.
"Second and third tier [ASIC] vendors are going out of manufacturing. They need to get out quickly and into design services because they can't afford to develop the process. Their strategies had better have shifted within the next five years or they're going to be out of the market," he s aid.
And then there was one?
Lewis indicated that while top-ten semiconductor companies would have the necessary economies of scale to prosper, any ASIC vendor smaller than an LSI Logic would probably be doomed.
"Single-chip systems really change the dynamics of the industry . . . there's a trend towards commodity markets for cell phones and personal computers," Lewis said.
Lewis portrayed the future as a tension between commodity markets, where economies of scale are all-important and there are few winners, and new niche markets where innovation rather than size is important and there can be many winners. Nonetheless the massive size of some of these maturing commodity markets, particularly the mobile communications market means the move to commodity is the stronger pole, Lewis said.
By 2004, Dataquest expects the digital cell phone to be clearly the largest ASIC market and worth about $22.5-billion and roughly five times larger than the next largest segment, hard disk drive applications.
Lewis said application specific standard product (ASSP) design starts will roughly equal ASIC design starts in 2002, at 7000 and ASSP design starts taking a lead about 7,500 compared to 6,500 ASIC starts in 2003.
In commodity markets economies of scale count, there is little differentiation in features, and there are only few winners. "Are we heading towards the automotive market model where there are just six big vendors? No but we are moving in that direction," Lewis said.
Daya Nadamuni, senior EDA analyst, presented the trends Dataquest sees in the embedded software market. She said there is a blurring of distinctions as handheld computers require RTOS capabilities and protocol stacks to handle communication, and as cell phones take on Internet browser and personnal information management functions.
Nadamuni observed that software revenues increased 13 percent in 1999 but were accompanied by several mergers and acquisitions. The arrival of embedded Linux promp ted the formation of a number of startups and corporate realignments while, a second major trend was the return of Microsoft with its CE operating system after being rebuffed in 1998 and 1999. "Microsoft is regrouping and refocusing on the 32-bit selected vertical markets," said Nadamuni.
Jim Tully, chief analyst at Dataquest's semiconductor group presented recent findings on the intellectual property (IP) or core-trading market. He estimated that the market, dominated by ARM, MIPS Technologies and Rambus, was worth about $417 million in 1999, up 36 percent from 1998, and was expected to grow at a similar pace in 2000.
He went on to examine the case for e-business in intellectual property (IP) cores, which in theory are a commodity that can be not only found and researched over the Internet but also delivered that way. Tully's conclusion was that e-business would not help much at present where most of the costs of acquiring IP are in the technical evaluation and technical support.
In Smith's pres entation, he also dwelt on the possibilities of Internet-based design, but observed that the size of the community was small compared with other over-the-Web activities.
"With less than half a million eyeballs there will probably be only one main design portal," said Smith. He then outlined alternatives where that design portal is operated by EDA vendors as an extension of the sales channel or by the engineers as a collaborative work environment.
Smith seemed to favor the latter when he said it was a natural inclination of the design engineer to collaborate. "You can argue the must successful thing in the Internet area is ESNUG [electronic Synopsys user group moderated by John Cooley."