The five forces driving the semiconductor IP market

The five forces driving the semiconductor IP market

Why designers are relying on IP to help them cope with the growing complexities in their SoC designs

BY JOHN KOETER
Synopsys
Mountain View, CA
http://www.synopsys.com

The semiconductor IP market is expanding rapidly in size and importance, and is forecasted to grow two times faster than the semiconductor market as a whole over the next three years. Five forces are driving the use of third-party IP and changing the way IP providers work.

Fig. 1. Five forces are driving the use of third-party IP and changing the way IP providers work.

These forces — referred to as the “five Cs” (see Fig. 1 ) — help to explain the increasing importance of third-party IP, in addition to why designers are relying on IP to help them cope with the growing complexities in their SoC designs. The 5Cs are:

• Convergence . Today’s SoCs incorporate a great variety of functionality into a single device, with ever more IP blocks on a chip.

• Core vs. context . With lower growth in the semiconductor market, expense management is a priority. Companies are actively outsourcing IP development and other activities not core to their product differentiation.

• Complexity . Moore’s Law marches on. As process technologies continue to shrink, the gap between available silicon and designers’ ability to fill that space continues to widen. IP is critical to improving design productivity at deep-submicron processes.

• Consumer . Consumer applications demand highly integrated functionality and competitive features, such as performance and power, all at an aggressive price point. Market windows can’t be missed. All of these factors drive the need for proven IP to reduce risk and achieve schedules.

• Consolidation. There are two driving forces for consolidation. As IDMs implement a fab-lite or fabless strategy, the number of fabs and processes is consolidating. By providing access to silicon-proven, high-quality IP during the foundry process, IDMs are able to increasingly outsource IP. The growth in complexity of IP and the required attendant verification and validation are also driving IP vendors to consolidate because few can afford the required investment to produce high-quality IP.

Primarily driven by consumer applications, these factors have rapidly risen in importance with new cost pressure, schedule pressure, and risk avoidance. These pressures cause semiconductor companies to increasingly choose to outsource IP, turning previously captive internal developments into merchant opportunities.

Convergence calls for wider expertise

As multiple chips are now rolled into one, a single design team must deal with many different types of functionality and many different industry standards. A single chip for a set-top box, for example, must include many standards-based interfaces, such as USB 2.0, Ethernet, DDR, PCI Express, and HDMI. It also must include multiple CPUs, complex audio/video algorithms, and extensive software.

With many pieces of logic coming together on a single chip, it is difficult and often impractical to be an expert in all of the different domains. For standard interfaces, more designers are turning to reliable third-party providers because the IP encapsulates the necessary domain knowledge.

Clearly, designing every transistor on a complex SoC, such as a set-top-box chip, is no longer economically feasible. And perhaps more importantly, compressing market windows mean design teams don’t have the luxury of time in the schedule to develop the IP from scratch. To hit reasonable market windows, teams need high-quality, third-party IP that they can depend on to reduce their risk and speed their time to market.

Core focus cuts costs

The complexity of today’s chips drives companies to focus their resources on technology that clearly and sustainably elevates them above their competition—their core differentiators, as Geoffrey Moore puts it. For chip design, this strategy calls for a focus on core competencies, such as the chip architecture, software content, unique application knowledge, and collaboration with lead customers.

This strategy does not include the nondifferentiated portion of their design, such as the standard CPU, embedded memories, connectivity, and interface IP. While companies certainly can design the IP themselves, doing so puts their core differentiation at risk and can result in unexpected costs and schedule delays as internally developed IP is reused.

Often, an internal core is not fully documented or supported, as it was created for a fixed application configuration. As a result, applying it to new applications with new design teams can require a significant design and verification effort by the group picking up the IP. Of course, the modified core may also have bugs or not be fully compliant with relevant standards, nor be interoperable with other third-party chips.

As expense management continues to be one of the highest priorities for semiconductor companies, it is increasingly common for large and small companies to purchase IP. By outsourcing IP, fixed engineering expenses (salaries, hardware, and software) are converted into lower variable expenses. This enables more budget and resources to be applied to the chip’s core differentiation, while reducing schedule and expense risk by using proven, high-quality, well supported third-party IP cores.

Complexity compounds SoC design time

Moore’s Law is alive and well. Design teams continue to see a growing design gap between the gates available and designers’ ability to use them. High quality, third-party IP is an increasingly common solution for closing this design gap.

Since companies that design SoCs are using outsourced IP, the interesting question is: how much outsourced IP do they use? According to Semico research, the average number of outsourced IP blocks will double by 2010 to over fifty IP blocks per SoC.

Just like SoC complexity, IP core complexity has skyrocketed in recent years. A USB 2.0 Host is 23x as complex as a USB 1.1 device.

Similarly, an ARM 1176JZ-S core is 24x more complex than an ARM7TDMI-S core. With USB 3.0 around the corner and the need for high-performance processors to keep up with growing compute requirements, this trend shows no sign of slowing down.

Keeping up with the complexity of the IP is just one factor to consider in the make-versus-buy equation. Thorough verification across all application modes, certification and interoperability testing, comprehensive documentation, expert technical support, maintenance and enhancement of the IP are all core values to an IP provider—while they are context to the SoC design team.

Consumers drive us

The consumer market is intensifying the other forces at work in the IP market. The dynamics in the consumer market are much different than the B2B market. Market windows are often narrower, product lifecycles are shorter, and fad/fashion-driven products introduce less predictability and more risk. Consumer devices also drive convergence, while demanding low cost, aggressive schedules, and increasingly complex architectures and software.

According to Gartner/Dataquest, the consumer market is over 50% of today’s semiconductor industry. Consumer products also dominate IP usage. These factors increasingly drive companies to use proven IP. High-quality, well-supported, and reliable IP that saves time and minimizes design and schedule risk is, therefore, a strong incentive for SoC designers to use proven IP.

Careful choices needed

Consolidation affects the IP market in several areas: IP vendors are consolidating; traditional IDMs are moving to a fab-lite strategy using merchant foundries; and merchant foundries are consolidating process technologies, sometimes across multiple foundries.

According to Semico research, over the past 10 years the number of small IP vendors has decreased by 80 % while the number of large IP vendors has increased by over 100%. Since high quality is the key factor to achieving the benefits of third-party IP, customers demand high quality from their suppliers. This requires a substantial investment in design, verification, validation, certification, and interoperability testing. An established IP vendor with more customers can spread the expense of IP development across a larger customer base and, thus, can invest more in quality and support (see Fig. 2 ).

Customers are also demanding complete solutions for their IP needs from vendors. Established IP vendors have the resources to provide complete, integrated IP solutions further decreasing risk and time to market. For these reasons, IP suppliers have been steadily consolidating.

Fig. 2. Developing high-quality IP requires significant engineering investment across multiple areas.

Fab consolidation is also a significant trend in the rise of third-party IP. Traditionally, an IDM with a proprietary fab could tweak the process to support product differentiation and, thus, justify creation of both unique and standards-based IP.

In a world with more fab-lite and fabless semiconductor companies and fewer fabs, buying third-party IP, particularly hard IP, makes even more sense. The new reality of increased standardized processes allows IP vendors to provide silicon proven, characterized, certified IP for most of the available processes. With fewer fabs and processes, IP vendors can reliably provide all of the IP needed in a given process technology. ■

For more on semiconductor IP, visit http://www2.electronicproducts.com/Software.aspx.