EDA '07 forecast: strong, but watch the bumps

SANTA CRUZ, Calif. — The EDA industry grew faster than expected in 2006 and should have another good year in store, according to executives from large and small EDA companies. But a shakeout among design-for-manufacturability (DFM) startups could result in a bumpy ride, some warned.

Factors fueling EDA growth include a healthy semiconductor industry, an insatiable consumer electronics marketplace and a move to 65- and 45-nanometer IC fabrication technologies. As a result of this move, DFM tools and technologies have come into strong demand. There's also been slow but notable progress with electronic system-level (ESL) tools and methodologies.

At the start of 2006, most EDA executives were predicting a somewhat lackluster year with single-digit growth, following four years of essentially no growth. In October 2006, however, a Gartner Dataquest report predicted a growth surge in 2006 of 13.3 percent, resulting in a forecasted $4.5 billion worldwide EDA tool market. The primary driver, the report said, is investment in DFM-aware IC design tools.

The Gartner report predicted a lower 8.5 percent revenue growth in 2007. Stronger growth will resume in 2008 and 2009, the report said, spurred by investment in ESL tools.

Why was 2006 such a good year? "After going through a long cost management phase, the semiconductor industry is again focusing on technology differentiation and innovation," said Aart de Geus, CEO of Synopsys Inc. Further, he said, there's been an "increased aggressiveness" in moving to lower technology nodes, with customers transitioning to 65-nm designs more quickly than expected.

"It is clear to me that the EDA industry leaders have returned to innovation, and are offering end-to-end solutions for the enterprise," said Jim Miller, executive vice president for the product and technologies organization at Cadence Design Systems Inc. He noted the development of "solid" design flows for 65 nm and said that many customers are skipping 90 nm and going straight from 130 nm to 65 nm.

Design complexity and shrinking process nodes have forced a reassessment of design methodologies, resulting in EDA revenue growth, said Wally Rhines, Mentor Graphics Corp. CEO. "DFM and ESL led the way in 2006, as did the continuing rapid adoption of SystemVerilog," he said.

Rajeev Madhavan, CEO of Magma Design Automation Inc., also cited a rapid move to 65 nm, but wasn't as impressed with the year's EDA developments. "In 2006, the emphasis and focus of EDA was on DFM, behavioral synthesis, sequential verification. Curiously, there was plenty of noise with little to show in the way of products."

The big question is whether EDA growth is sustainable into 2007 and beyond. Rhines thinks so. "Rapid growth of 65-nm-and-below designs will continue to force new tool developments, and adoption of DFM and other design tools," he said. "For Mentor, software renewals for 2007 are looking very positive."

Signs that semiconductor R&D spending will continue to grow bodes well for EDA, said Cadence's Miller. "The current market segments and business drivers are strong," he said. Miller, however, predicted "considerable consolidation" in the DFM market.

Magma's Madhavan said that many of the venture-capital-funded ESL and DFM startups will close down in 2007 and 2008. "ESL product development will continue to be slow, and standalone DFM products will have a hard time surviving," he said. "Both will result in a large number of failures in EDA and cause the venture community to stay away from the industry until the next success."

DFM's mixed prospects
There are varying views on what will happen in the DFM market, where a number of startups are marketing point tools as the big EDA vendors add capabilities of their own. The big story of 2007, said Atul Sharan, president and CEO of DFM startup Clear Shape Technologies Inc., is "that the reports of DFM's pending death will be proven to be extremely premature and short-sighted."

Sharan pointed to ASML's December acquisition of Brion Technologies for $270 million as a sign of DFM's strong appeal. "DFM is clearly where the action has been and will continue to be," he said. "Having said that, not all DFM startups are going to make it."

"DFM as a design challenge is here and it is real," said Pravin Madhani, president and CEO of Sierra Design Automation Inc. But there will be winners and losers, he noted. "The DFM startups with differentiated or comprehensive solutions will survive, while most of them will perish," he said.

Large EDA companies are more cautious about acquisitions than they previously were, and overinflated valuations are a thing of the past, noted Roy Prasad, CEO of DFM startup Invarium Inc. "Startups will also have to look at mergers among themselves to achieve higher critical mass and address larger flow dynamics, not just point solutions," he said.

Prashant Maniar, chief strategy officer at Stratosphere Solutions Inc., distinguished between DFM, which he views as a "process-centric, deterministic simulation play," and design-for-yield (DFY), which he sees as a "front-end play." While DFM startups will face an uphill battle in scaling business beyond early-adopter customers, he said, designers will gravitate toward an "actionable" DFY flow.

ESL inches along
Observers have been predicting for years that ESL will represent the next big wave in EDA, but for 2006 and 2007, even advocates see slow going. In 2006, said John Sanguinetti, CTO of Forte Design Systems, "ESL became a mainstream focus of EDA activity, but it has not matured into a mainstream part of the design flow. This was a gradual process, and I don't think there were any dramatic milestones."

"In 2006, there were examples of ESL tool deployment, but also signs of an immature market," said Tom Sandoval, CEO of ESL verification startup Calypto Design Systems Inc. "There are gaps in the ESL methodology." The message from customers, he said, is clear: "Provide complete ESL solutions or we will build them ourselves."

The most important ESL development in 2006, said A.K. Kalekos, vice president of marketing and business development at CoWare Inc., was the emergence of virtual platforms for software development. "It may very well be the killer application for ESL design, and may be the catalyst for changing how electronic products are designed," he said.

In May 2006, Synopsys Inc. bought Virtio, a provider of virtual-prototyping tools that let engineers develop software using models of system hardware. Creating virtual prototypes can save a great deal of money by reducing test runs, said Synopsys' de Geus. "The increased quantity and criticality of embedded software will drive the need to enable software validation on the intended hardware devices much earlier," he said.

One driver for ESL is the increasing need for low-power design, which is best addressed at higher levels of abstraction. "Design-for-power is the new mantra," proclaimed Vic Kulkarni, president and CEO of Sequence Design Inc. "Our work with ESL partners is revealing how powerful microarchitectural design exploration can be."

A custom approach
Not all IC design is standard cell-based digital design, and there was a renewed focus on analog and custom IC design in 2006. The big development there, said James Spoto, CEO of Applied Wave Research, is the emergence of the OpenAccess database. In 2007, he said, look for a push to extend OpenAccess to include interoperable parameterized cells (p-cells), standard technology files and parameter-mapping standards.

"2007 will be the year that analog design tools will begin to close the gap—if only slightly—with the digital world," said Andrew Levy, vice president of marketing at startup Lynguent Inc. "Although the killer app—analog synthesis—is still off in the distance, a number of companies are biting off more focused and manageable chunks of the analog/mixed-signal flow."

Ken Roberts, CEO of Pulsic Ltd., predicted that 2007 will bring about a "fresh market emerging in the automation of custom design flows" for analog and custom digital design. RTL-to-GDSII flows for large digital ASICs have become "commodity solutions," he said, and will consequently show little growth in the new year.