First of two parts: Lines are blurring between segments as upside of some markets taper down and new ones take off. Pain, efficiency and expertise define opportunities.
Look at the top line numbers provided by the EDA industry consortium (EDAC) and it appears as if the industry is doing well. In 2010, revenue was $5.285 billion. That number increased to $6.218 billion in 2011, and again to $6.529 billion in 2012, a 9.5% annual growth rate that would satisfy most investors. But the numbers do not tell the whole story. There is an interesting divide growing between the top three EDA companies and those from neighboring industries.
Top line year-over-year growth for Q3 2013 was 6.8%, a little lower than the previous EDAC numbers but quite explainable given that Q4 is usually the biggest quarter for EDA. Within that growth, the numbers are split into three primary areas—EDA revenue, semiconductor IP revenue and services. EDA Revenue grew at 4.2%, semiconductor IP (SIP) at 12.7%, and services at 10.4%.
These numbers are not an aberration – this has been the normal state of things for the past decade. Core EDA shows less robust growth, which is why the industry gets little respect on Wall Street and receives little in the way of venture capital. On the plus side, this growth has been a lot more stable than most industries.
This does not mean that high growth areas cannot be found within EDA. Physical design tools tackling the latest problems at new geometry nodes are in high demand. “The problem EDA companies have is that the number of designs starts has steadily declined as process geometries got smaller,” says Jonah McLeod, corporate marketing communications director at IP vendor Kilopass. “EDA companies grow their business by licensing to design teams starting new designs or by charging a premium for tools to design at smaller geometries. Neither option is viable.”
While these tools are on the rise, others are in decline. Many agree that this is a natural progression for the industry. “New tools can command high prices, old ones get commoditized,” says Aveek Sarkar, vice president of product engineering and customer support at Ansys-Apache. “We have to adapt and solve new problems.”
But some see a brighter future. “We have already started to see design starts increase,” says Craig Cochran, vice president of corporate marketing at Cadence. “This is in part due to improving economic conditions, new places doing design such as China and India, new vertically focused companies such as Google and Facebook, and there will be another kicker when we see trends like the Internet of Things (IoT) gain more traction.”
Serge Leef, vice president of new ventures and general manager of the System-Level Engineering Division at Mentor Graphics, lays out a framework for us. “In any market you can sell current tools to current customers, current tools to new customers and sell new tools to current customers. The dynamic of each are fairly well understood.”
Current Tools to Current Customers
Leef agrees that the strategy of attempting to sell more of the same to the same few people is not very exciting and you would not expect to see much growth.
“While transistor count growth is impressive, most of the growth is actually in memory and mostly SRAM,” says Charlie Cheng, CEO of Kilopass. “The actual design ‘complexity’ is not in the transistor count of the logic, where EDA companies in the past have focused their efforts.”
If we subtract out the SRAM content of typical chips, it is believed that the EDA business probably grows roughly in-line with logic transistor growth. Cheng further predicts that “it’s plausible to posit that EDA has reached an asymptote on its value inside the R&D expense of a semiconductor company’s P/L.”
New Tools to Current Customers
Solving the challenges associated with new nodes, new devices and the inherent yield problems associated with the latest generation is a major investment area for all of the large EDA companies. In addition, emulation is growing in an attempt to keep the functional verification methodology on track and formal methodologies are making progress. ESL, while small, is making inroads and should continue to provide grow for the foreseeable future. At the same time, other product areas are in decline. PCB is one of the most mature areas within EDA and once was the largest revenue segment. Today it represents less than 10% of total revenue.
“There has been a fairly constant relationship between EDA and the semiconductor industry,” notes Sarkar. “The next wave of growth is likely to come from companies that help them manage complexity and to design more efficiently. There will also be rewards for companies that enable their customers to add value.”
But even with new tools there can be differences in the way that the problem is attacked. Most of the time large EDA companies make linear extensions to their existing tools. This enables them to get to market faster and to minimize the risk. When they depart from their traditional strategy and work on a new solution from the ground up, it can provide a significant boost in their market share, but the risk of failure is higher if they cannot support all of the requirements of their existing customers.
“An essential element to growing the scope of a business is understanding the space,” says Lianfeng Yang, vice president of marketing at ProPlus Design Solutions. The company saw a way in which existing technologies and expertise could be used to attack a new market segment.
New Tools to New Customers
If Kilopass’ Cheng is right, any gains made from developing new tools for existing customers will result in decreasing margins for the existing tools, such that the combination of the two will revert back to the same total revenue. To break free of this cycle the industry has to look at expanding into new markets with new products, and that means ensuring that it is a new pool of money they are going after rather than fighting for a larger share of the existing pot. The industry is doing this in two independent ways. The first is associated with IP and the second with adjacent markets.
“EDA has always focused on enabling efficient implementation and verification of complex designs,” states John Koeter, vice president of marketing for Synopsys’ Solutions Group. “One of the most efficient ways to implement designs is to use pre-designed, pre-verified blocks.”
In the last quarter of 2013, Synopsys reported that IP now represents 25% of its revenue and is growing at about 25% per year. At this rate it will not be long before Synopsys is an IP company with some tools on the side.
“From transaction-level models of IP for software development, to verification IP for protocol and system verification, to IP assembly tools and IP subsystems, it’s natural for customers to turn to their EDA vendors for not only the tools but also the IP needed for their specific designs,” says Koeter.
Moreover, they’ve been able to get paid well for developing that IP. “The EDA industry has done a surprisingly good job of acquiring IP and establishing a separate business model from EDA tools” observes Tom Anderson, vice president of marketing at Breker Verification Systems. “For the most part, EDA companies have been able to keep IP out of the ‘all-you-can-eat’ deals that have done so much damage to the industry.”
Part of this is good business practices from companies that have mastered M&A over the years. “EDA companies have taken a proven formula of waiting for the market to validate the needs for IP and then acquiring the IP company to scale the revenue,” says Cheng, “The very nature of these acquisitions is that EDA companies are willing to pay a premium after the product and market risks have been eliminated, and only focus on scaling the revenue.”
The line between EDA and IP has certainly been blurring, especially when highly configurable IP is considered. While the EDA industry is happy to include IP into their revenue numbers, the IP industry does not quite see it the same way. Several of the largest IP companies, including ARM and Rambus, are not EDAC members and have expressed no interest in being seen as part of the EDA industry.
“We prefer to be seen as an IP company,” says Grant Pierce, president and CEO of Sonics. He explains why he see the two functions are different and in addition says, “There are differences based on who makes the decision to use the technology. In EDA there is often a centralized manager for all EDA tools. While he takes input from many people, he is primarily a purchasing person and will have all of the EDA people come in and compete with the others. The people who are fundamentally making decisions about IP are the chip architects or VPs of engineering.”
While Pierce may not want to be seen as an EDA company he does admit, “We welcome that EDA companies are interested in buying IP companies.”
Some EDA companies see this as the natural path forward. Others prefer to go after new opportunities. Leef points out that “EDA traditionally focused on parts of the electronic design where the problems were more acute and there were no alternatives.” He sees other industries where the methodologies being used are no longer viable and, this creates new opportunities. Leef provides automotive as an example. “The complexity of these systems now requires advanced design techniques. Just as in the 80s you wouldn’t think of taping out a chip without full chip simulation or DRC, it is getting to the point where the electronics in a car cannot be designed without some degree of analysis and simulation.”
Leef believes that their methodologies need to be updated. “Continuing to do the same thing they have been doing for the past 100 years is not viable. This creates an opportunity to leverage EDA technology.”
Automotive is a vertical segment that interests all of the major EDA companies. Koeter, notes that “in automotive there are safety-critical systems that need to be thoroughly vetted before they will make it into production cars.” He sees this as an opportunity for established IP providers. “This will naturally lead to a longer time to a return on the initial investment, making it more difficult for small IP companies to be successful in automotive. This industry also requires strong knowledge of relevant standards”
In addition to going after vertical market segments, primarily with products based on existing technologies, another possibility is within completely different tool spaces, such as embedded systems, or where two markets are coming together. “At Ansys, we are not just looking at electronics,” explains Sarkar. “We are looking at complete systems. The lines between electrical and mechanical are blurring. Many domains are coming closer together and we have to look at how they interact with each other. That creates opportunities. Automotive is shifting towards electronics with increasing amounts of it in every car. It is not that EDA is getting closer to automotive.”
Leef explains the impact of optimizing a car, such that you can eliminate a single node from the network, would save $50 on each car. “Their existing technologies are unable to find this type of optimization, but the EDA industry has algorithms developed that can attack this problem today.”
In the second part of this series, we will examine where each of the major EDA players are placing their bets and the reaction to this from companies who are looking from the outside and seeing the same opportunities.
To view part two of this series, click here.