The design-services business has been around since the day the first disgruntled radio engineer unplugged his soldering iron, packed up his vacuum-tube handbooks and stalked off to start him a consultin' shop. For generations-of both engineers and technology-design services has followed that one simple model: individual practitioners, usually with special skills, selling their services to design teams.
Today, in a large part of the business, not much has changed. Much of the design-services market consists of individuals or small teams with special skills. These little islands of expertise generally engage with a design team much the way the local fire department engages with buildings-only when something has gone very, very wrong. But unlike the fire brigade, which rarely gets to stay for coffee and donuts, design-services practitioners often build lasting relationships with the teams they help. Sometimes these relationships turn into permanent des ign partnerships, and sometimes into full-time employment.
If there is a difference in this part of the market, it is the range of special skills that various design-service practitioners can bring to the design team. A design shop might be founded on knowledge of a particular architecture, such as the Amba bus. It might be built on the uncanny skills of one individual at tuning Verilog for high-speed data paths, or at achieving timing closure, or at packing the most stuff into a given family of FPGAs. Today there are also specialists for a whole range of emerging areas, such as formal verification or behavioral modeling. There are specialists dedicated to particular tools and particular application areas.
But the groundswell of change right now doesn't involve the individual practitioners at all: In fact, it is a growing shift away from the small shop-the industrialization of the design-services business-that characterizes the market today.
The complexity monster
Consolidation can only happen in a market that is becoming more organized. So to understand what is going on in design services, we need to study the trends in the market. We will in fact find both rapid growth and increasing organization: More design-services contracts are being signed, and the old firefighting nature of the market is changing. Increasingly, today's design-services consumers are asking for similar kinds of things, in predictable ways. Several forces are driving these trends.
The most obvious is our old enemy complexity. As designs get harder, it is more and more unlikely that a given design team will have all the skills necessary to finish. Or, as David Hardman, vice president for professional services at Synopsys, put it, "There's a discontinuity in what it takes to finish a design-it's not just about getting the RTL right any more."
Design teams are finding that they lack the skills to use new tools that have become necessary to deep-submicron flows. Often, a team's need to integrate a new tool without losing time will result in a design-services contract: Come on in and use this thing for us, and show us how you are doing it. It's important to note that this is a preplanned design service, not a firefighting expedition. Some large design houses claim that much of their business with advanced design teams is for specific services within an existing flow.
Another key driving force is related, but different. OEMs-and even some fabless semiconductor companies-that have traditionally relied on ASIC vendor relationships are moving to a customer-owned tooling (COT) flow. It's probably not too strong, based on ISD's research, to call this a movement. But deciding on a COT flow and licensing the right tools to take a design all the way to tapeout are a whole lot different than finding engineers capable of doing physical design and verification, or managers who have personal relationships at foundries. Even though hiring is a lot easier than it was in 2000, you don't find people like that in your in-basket.
Hence, there is a rapidly growing market for design-services vendors that can act as a surrogate ASIC company-working with a team starting at behavioral design, but really getting engaged during synthesis, and taking over almost completely during the back end. Such contracts are becoming common between large OEMs and design-services companies, and are often brokered by foundries as a service to the OEM.
If OEMs are being drawn beyond their traditional design expertise, they are also being drawn beyond their application knowledge. It's fine to be in the mobile-computing business, for instance, but these days that implies the ability to speak 802.11a, Bluetooth, related European and Japanese standards, and maybe MPEG-4 compression-areas that are applications in their own right. Convergence-to summon up a much-abused and discredited word-is pulling design teams into whole application areas beyond their comfort zones.
And that is creating yet another market for design services, based not on specific parts of the design flow but upon applications knowledge. Synopsys Professional Services, for example, maintains several local application expertise centers, including design centers devoted to wireless and to broadband. LSI Logic's Mint Technology group-yes, a design-services company within an ASIC vendor-has deep experience in embedding ARM cores.
Such groups focused on application areas tend to engage with their clients much differently than service teams focused on a particular design phase or problem. To begin with, they engage early-often in the behavioral design, or even the planning stages. At this point a little bit of hard-won experience can save major problems downstream.
Additionally, such relationships usually blossom during verification. In many application areas, the real secret sauce, to be frank about it, isn't the RTL. It's the testbench. The main contribution of the design-services team may be in the testbench code they bring with them.
"The fun ctional specification is really the key," observed Mint's Mark Stamvick. "If it is complete, the RTL is just a rewrite-maybe 15 percent of the job. And it's only that big a portion because the engineers insist on polishing as they go along. The next big job is identifying the key cases and creating a test plan."
Design-services shops may also play the role that was originally envisioned for intellectual-property providers. The notion of plug-and-play IP is so thoroughly discredited at this point that many question whether there is any such thing as a pure-play IP company. One answer to that question is yes-but it may be indistinguishable from a design-services house. "In many accounts we have become the IP provider of choice for all those complex blocks that don't differentiate the OEM's end product," said Paul Russert, director of business development at Tality. "For instance, we have baseband and RF blocks to use off the shelf in Bluetooth applications. Those blocks are just a matter of complying with the standard-they aren't the client's secret sauce."
Finally, a more complex role for design-services firms is emerging, more as a trend than as a fact of life today. That is the role of methodology provider. Often design managers perceive their need for help in terms of worries about verification, about design closure, or just the uneasy feeling that the flow isn't under control. These days, even teams in large OEMs often can't justify-or find-an experienced methodology manager. So they turn to design-services companies, not for any point assistance but to help with the management of the overall flow.
A case for bigness
All of these trends are driving growth, but less obviously they are also pushing the industry toward consolidation. Not that the individual practitioner with unique skills is going away. According to those gurus with whom we spoke, exhaustion is more of a danger than boredom today, especially in problem hot spots.
But the center of the design-services industry is unque stionably moving away from the genius with a shingle and toward the big, corporate design house. Tality, for example, boasts 800 engineers-only half in North America-and pocketed $200 million in revenue in 2000. Synopsys Professional Services is around 400 engineers and hiring. Even ventures that started out as quite application-specific small shops are either growing or being absorbed. Zaiq, for example, is a shop with specific expertise in verification of 10-Gbit and 40-Gbit communications cores. It is also a 200-engineer design-services company broadening its industry base.
All of these trends tend to bias the industry toward bigness, according to at least the larger players. The shift away from firefighting and toward partnerships demands not just a brilliant individual, but a sizable team of players who are used to working together. Such a team needs a lot of infrastructure-including server farms and tool licenses-to support them. And that infrastructure, in turn, is more affordable if it's amortize d across several teams.
The nature of the help is changing as well. "Increasingly, clients aren't looking for a contractor to solve a specific problem," said Zaiq's chairman, Rich McAndrew. "They are looking for a partner who will see them all the way through this product's life cycle, and use that experience to help even more on the next one." That's a call for stability, and for size. And it's as good a statement as any of the trend that is gradually transforming the design-services business into one dominated by really large players.