Executive Outlook

12 out of ’12: A dozen of the industry’s leaders sound off on the big issues facing the semiconductor design industry over the next 12 to 24 months.

popularity

By Ed Sperling
The view from the top of companies is a like a high-level of abstraction for viewing the industry. While engineers get caught up in individual projects, or pieces of projects, CEOs and CTOs tend to see things from a much broader perspective.

So what do they see as the big issues and developments over the next 12 to 24 months? System-Level Design asked industry leaders that question. What follows are their answers:

Wally Rhines, chairman and CEO of Mentor Graphics
“The big issue is adjusting to the changes in semiconductor consumption, from computer-focused to communications-focused customers. And as we move to 20nm and ‘14/16nm’, the challenge will be maintaining the traditional 30% per year reduction in cost per transistor with less help from wafer prices and more impact from design and 3D structures. We’re also moving from simulation-dominated verification to a simulation/emulation verification environment for leading-edge designs.”

Aart de Geus, chairman and co-CEO of Synopsys
“We’re about to see a fabulous transition to 20nm and finFETs. FinFETs are a big step to lowering power. There are a lot of things you can do with this technology, and we’re absolutely seeing a race with finFETs at the center. That’s going to open up a new wave of development around performance and power, but it also will make it difficult for all parties to be successful. This is an immense race among the big CapEx spenders. And we’re likely to see a number of unexpected applications coming out of this.”

Lip-Bu Tan, president and CEO of Cadence
“We’re going to see more and more customer concentration, and more system integration. It’s interesting to see the foundry model changing to adjust to this, as well as emerging growth in China and India. We’re also seeing fewer and fewer startups because it’s harder to develop chips these days. If you think about just the mask costs at 28nm and 16nm, it’s very high. That means the chips developed at those nodes will have to be very high quality. But we also have to start adding innovation back into the industry. If you add up the cash inside of semiconductor companies it’s about $400 billion. That’s a lot of cash in an industry where everyone is doing incremental changes but no one is changing the game. We have to move in a different direction over the next year or two or no one will grow.”

Lucio Lanza, managing director of Lanza techVentures
“The big shift is that designers need to realize the main driver on the evolution of process nodes. It used to be the size of memory and performance. Now it’s power and performance. People don’t switch anymore from one node to the next because of performance. The other huge challenge is the virtual IDM model. This is a big change.”

Naveed Sherwani, president and CEO of Open-Silicon “We’re seeing a lot of indecision for the first time in memory. Companies are backing out of building chips and carrying that over into the next year. That doesn’t mean it’s negative on growth. Networking and cloud silicon will continue to go up, and there are a lot of people adopting 28nm. We’re going to see a big push to 3D-IC when Micron releases its 3D memory, too. Companies are going to look at it, realize its benefits, and buy it. This is going to offer a lot of improvement in system-level performance, and it’s going to provide power and space savings. For big data centers, 3D memory will become very important because it allows them to decrease the size and the power of the data centers.”

Scott McGregor, president and CEO of Broadcom
“The big trends facing the semiconductor industry over the next year or two are the ability meet demands of some important market opportunities, while at the same time continuing to make the most of new process technologies. There are significant opportunities for semiconductor companies being driven by 5G WiFi (or 802.11ac wireless), NFC, Ultra HD television, LTE, the development of faster networks (40G and 100G for data centers and office networks approaching closing in on one gigabit per second) as well as automotive Ethernet and connectivity. At the same time that these areas are poised for significant growth, many emerging countries are accelerating their own communications build outs. While the opportunities in these countries is enormous, they need products focused on their own requirements based on current process technologies. The industry’s ability to meet the demands of these areas will determine an overall level of growth and could help speed an overall recovery.”

Kathryn Kranen, president and CEO of Jasper Design Automation “The big issue we’re hearing from our customers—and the big opportunity—is security. They’re needing to prove that a device is immune from attack. This is verification of illegal traffic. We’re also seen a big focus on cache coherence and architectural decisions. Companies are moving coders up to the architectural stage. And we’re seeing an increasing focus on low power and the need to implement this correctly. The question for a lot of designs is whether low power breaks functionality.”

Jack Harding, chairman and CEO of eSilicon “The big issues are all related to process going forward. The market has accepted 28nm, but the cost per transistor is going up at 20nm. That will force companies to rethink value. FD-SOI is a good alternative for some applications. 2.5D will be a big trend, too. Some companies are saying they’re not going to 20nm across the board. They’re doing it selectively. They’re also going to be buying packaging alternatives. The value proposition will be based on total cost, not chip cost. At 20nm, you’re probably looking at a minimum of $4 million in layout costs and it will require 15 Ph.D’s.”

Charlie Janac, chairman and CEO of Arteris “The big question that needs to be resolved in the phone market is who wins the middle of the market. The high end is already locked up by Apple and Samsung. But what will happen in the midrange and the low-end of the market where there is a huge volume opportunity? Will it be Intel, MediaTek, Qualcomm, or some other company? The other big question is how does the digital TV industry get re-made? Will there be a visionary who can take this market in a new direction rather than just increasing the size of the screen or turning it into a giant tablet?”

Chris Rowen, CTO of Tensilica “The big issue we’re seeing is an enormous shortage of software, which is the underlying technical capability of the platform. The user experience is not about a higher frame rate or larger glass. It’s about marrying content to interactivity. That’s particularly true in the digital TV space. It’s really about a transition to an immersive interactive experience. It’s about getting it to work well and interact with the Web more seamlessly. Right now, you don’t need to be inside the story. But you do have to deliver services around entertainment.”

Prakash Narain, President and CEO, Real Intent “On-chip complexity will see a major jump in the next silicon node. The 20nm node may have 14nm transistors, or semi companies may skip 20nm and go straight to 14nm. Either way, there will be significantly more transistors in the next node than in the typical transition. Impact will be felt on EDA tools across the board. The largest impact will be on verification tools that don’t have the luxury of a sliding scale of Quality-of-Result (QoR).  Today, we are seeing 450M gate SoC designs for Clock Domain Crossing (CDC) verification. With the new nodes and the demand for ever-greater feature sets and performance by consumers, 1G gate designs will be the top end in 12 to 24 months. The challenge will be to scale verification tools to handle these designs and give signoff-accurate results in hours and not days or weeks.”

Grant Pierce, president and CEO of Sonics “There may never be a more exciting time in our industry. We have a sea change taking place in the market for personal computers, smartphones, tablets, servers and other connected devices. Major semiconductor franchises are giving way to new players serving new markets in new geographies. It’s a new vibrant connected world of powerful devices, location-aware always available. And yet much of the song remains the same, with customers planning 2013 chips driving higher SoC complexity to support the end user demands for better performance and lower energy usage. Architects clamor for better analysis tools and system IP that guarantees their design intent all the way from conception though silicon. New multi-core system requirements for coherency and security stress internal solutions beyond their limits. We could not ask for a better New Year!”