Sorry, you need to enable JavaScript to visit this website.

Innovate beyond silicon

Recently, Marc Andreessen projected that within 20 years, ambient computing will be a reality. Every physical item—every table, every wall, every surface—will be sensing, computing and connecting. Such ubiquity will require a new generation of semiconductor devices, because semiconductor manufacturing technology is hitting a wall. The laws of physics and economics are imposing limitations on the pace and cost-benefit tradeoff of further transistor scaling. The industry is finding it challenging to make chips smaller and faster, which is the essence of Moore’s law.

More than moore

A new roadmap, called More than Moore (MtM), is gaining momentum. It encourages chip and hardware developers to think beyond the complementary metal-oxide semiconductor (CMOS) trajectory and to consider other technologies in their product development roadmaps. MtM is particularly relevant for IoT application development, given the inclusion of analog sensing technologies and the low-power, low-latency requirements of connected things. 

For example, micro-electro-mechanical systems (MEMS)—such as gyroscopes and pressure and motion functions, among others—are critical for virtually all IoT applications. Also, new and innovative packaging technologies, such as layered silicon, will reduce the size of devices and the time circuits need to wait for instructions. 

To take full advantage of MtM, the R&D organization should watch three value-seeking opportunities: chip specialization, algorithms and ecosystems. 

here is my title

Innovate beyond silicon

Find value in specialized chips

Semiconductor innovation will continue to be a building block for the Internet of Things. Take, for example, Intel’s Curie chip, which could be the next ‘Intel Inside’ product breakthrough. Curie is Intel’s answer to the consumer wearables market, a button-sized SiP that includes a processor, flash memory, sensors and Bluetooth low-energy connectivity. Company CEO Brian Krzanich said he thinks Intel can sell hundreds of millions of Curie chips in the athletics sector alone. 

Increasingly, the focus is shifting from high-volume ASIC and SoC market opportunities—which were the traditional focus of previous digital eras—to fashioning specialized chips from existing design libraries for niche applications. In 2015, for example, ARM introduced a high-performance 64-bit CPU core, called the Cortex-A35, that is capable of running advanced operating systems such as Google Android, Linux and Microsoft Windows. The company followed up with a 32-bit stripped-down version that targets low-power IoT devices like wearables. 

Similarly, Qualcomm recently announced a power efficient version of its Snapdragon SoC for wearables that combines Wi-Fi, LTE and an integrated sensor hub for richer algorithms. Adesto Technologies released a low-power conductive bridging RAM (CBRAM) chip for battery-operated or energy-harvesting devices that temporarily stores data before transmitting it. Niche markets for these products include medical devices, industrial machinery and HVAC systems. 

Complementing chip specialization is a need to ensure greater security. Anything that is connected can be reached, and potentially breached. Manufacturers of IoT devices and systems are right to demand that chipmakers ensure high security standards for semiconductors. A major security breach, such as the widely reported 2015 Jeep Cherokee hack, could slow down IoT adoption. NXP’s latest chip, a low-power 64-bit ARM-based processor, comes with built-in networking-grade security. 

Finally, a new breed of merchant ASICs is driving low-cost specialization. These chip suppliers, including Arista, help networking gear makers design their own chips using off-the-shelf silicon from Intel, Cavium and Broadcom. Downstream in the value chain, service providers, network equipment providers and enterprises are loading software—primarily open source—onto the merchant silicon-built hardware. 



Find value in algorithms

In the 1990s, Rolls-Royce changed its business model from selling jet engines to selling miles and competing on outcomes. Algorithms ingested streams of data from a multitude of sensors to forecast the likelihood of wear and tear, to anticipate the maintenance of individual engine components and to calculate the cost per mile for the customer. In doing so, the company transformed its engagement with customers from a product relationship to a service relationship.
Although embedded systems are nothing new for industrial equipment makers, they are becoming a major new source of value in the transformation of products into smart services. Companies that think they sell hardware should shift to selling services, which will require developing new business models, talent and partnerships. 

The value creation opportunity in software lies squarely in the realm of artificial intelligence and proprietary algorithms. Advanced Driver Assistance Systems are a good example of algorithms that learn, reason, experiment and act with a goal in mind. Apple’s HomeKit, ARM’s mbed, Google’s Brillo and Samsung’s Artik exemplify pioneering IoT platforms that take advantage of machine learning.

Innovate Beyond Silicon

Power of algorithms for managing costs

A recent Accenture study captured the power of algorithms for managing costs, improving the customer experience and increasing revenue. The company analyzed 30 pilot projects at early technology adopters that were using machine learning algorithms, including anomaly detection, natural language, predictive analytics and visual sensing. In one pilot, a consumer food company machine-reengineered the delivery of its products using a collision avoidance system with an intelligent vision sensor. The system scans the road while applying computer-vision algorithms, and then sends alerts to the driver. The project resulted in a marked reduction in accidents and delays, with implications for cost, revenue and on-time delivery.

Artificial intelligence makes up a significant strategic push for industry pioneers. For instance, SpaceX and Tesla founder Elon Musk, LinkedIn’s co-founder Reid Hoffman and corporate backer Amazon, among others, have contributed to a $1 billion investment in a new Artificial Intelligence Research Center. The goal, Musk said, is to create “artificial general intelligence,” where a machine would be capable of performing any intellectual task that a human can perform.

Innovate beyond silicon

Find value in ecosystems

Over the last 10 years, design-led companies have maintained a significant stock market advantage, outperforming the S&P by 219% in 2014, according to an annual report by the Design Management Institute. The report also noted the emergence of a new set of potential design leaders, including Amazon, GE, Google, Honeywell and SAP. Honeywell, for example, collaborated with third-party design firms to develop Lyric, a connected temperature control system that competes against Google’s Nest automatic thermostat Lyric connects to a user’s smartphone and uses geofencing to automatically set the home’s temperature. By tracking the phone’s location, Lyric knows when to turn down the heating or cooling; it then brings the home back to a set temperature when the user returns


  1. Leverage Moore’s law but exploit complementary technologies to power IoT developments. 
  2. Develop expertise in algorithms for IoT applications that deliver compelling services. 
  3. Build and participate in the development ecosystem to seed and grow new markets.