Over the past few years, being closer to the consumers has become extremely important when it comes to electronic product ecosystem. This provides the knowledge of end-user requirements to product managers and helps them decide on products with high success probability. Today, OEMs (and in some cases even service providers) know exactly what they need in a system-on-chip design. The IP, OS and other ASIC houses are slowly moving into a model where they can leverage the knowledge of multiple semi vendors and effectively amortize their costs. This industry structure is likely to address the key challenge: rising costs of semiconductor development.
Taking an example from cellular service: since its mass market introduction in the ‘90s the cost of voice calls has constantly declined and the consumers have been demanding more enhanced features in their mobile phones. From a point where SMS was a key value added service, today we are looking at video, instant messengers, GPS, Wi-Fi based browsers, music players, etc, all built into a single unit that is smaller than the basic cellphone 10 years ago. Mobile phones have essentially doubled up as a multimedia device and the applications on a mobile phone are still evolving.
However, these additional features come at a cost. Our efforts towards incorporating high-definition video to the list of features in a communication device are likely push the complexity to the next level. While engineering teams in the ecosystem today certainly have the capability to execute such designs, the key question is - Can business produce such designs and stay profitable?
Cost challenges
The key cost of such an IC is likely to be alarming, looking at recent trends. At 0.13m, the average design had 9 million logic gates and cost about $9 million to design. However as we moved to 90 nanometers, the average gate count increased by 80% and design cost almost doubled.
Today, the in-vogue 45-nanometer design costs have reached $70 million per design making it practically impossible for a small firm to tape-out a design on its own.
This meteoric rise of costs (8x) from 2004 to 2008 has called for significant changes in the way semiconductor industry is aligned today. It also tells us that the tolerance to failure is negligible.
To have a successful design it is important to understand end-consumer requirements.
Co-innovation in R&D
A few companies recently not only shed the ‘not-invented-here’ model but altogether realigned the way they do business to best leverage the ‘co-innovation model’. The trend towards more feature rich devices to cater to consumer markets has forced companies to look at the way companies can use their internal R&D team better.
Earlier each business group within an organization ran its R&D almost entirely on its own. Each had its own market facing teams, R&D teams and the supply chain teams. This, of course, led to a huge overhead and duplication of platform related efforts.
In order to minimize overheads and redundancies, the technology or the R&D groups within most firms have today been realigned as a single entity. This team ensures the technical decision-making is not only of maximum benefit to the organization but also is what consumers need.
Recently we are seeing this phenomenon going beyond the organizational boundaries. Firms are also making parts of their technology (IP cores, peripheral blocks, etc.) available to other vendors. All this is likely to reduce the overall cost of development of products and ensure higher ROI. STM-NXP merging their wireless business into a JV can be seen as a positive move in this direction.
Nothing new for Indian firms
This trend of co-innovation is deeply rooted in business models of leading product engineering firms in India.
The focus from firms toward consistent implementation from engineering teams has allowed partners to introduce both new products and product variants into the market within schedule time and costs. Product engineering houses have also been consistently working toward reducing cost and development complexity by maximizing reuse of IP and design components
This has enabled the VLSI industry in India to reach $766 million in 2007. This number is expected to grow at a healthy cumulative growth rate of 21.7% over the next three years with increasing share of projects coming from IP development and chip testing.
Indian design houses continue to see an increase in complete (spec to tape-out) type of designs. This has been a remarkable change in the total number of designs that are completely being carried out of India. This is a far cry from the early days of the industry (‘90s) and implies that the confidence level in Indian engineering teams is rising.
Co-innovation is the future
As each product introduced in the market needs to be exactly what the consumers are looking for, business teams need to be sure that the volumes required to breakeven would be reached in the time desired. To create such products OEMs are expecting complete solutions to increasingly come from semiconductor firms. The semiconductor firm in this case is responsible for utilizing the entire ecosystem to deliver the product on time with the required features.
With the expertise of Indian design houses in leveraging the ecosystem, providing quick turnaround and lowest cost of ownership over the past decade, their position as the overall global design hub is likely to grow stronger.
You can reach Anuj Sharma at Sharma.anuj@wipro.com
