IoT’s 3 Big Demands On The Semiconductor Industry

We’ve moved beyond dropped calls. The next phase is much more interesting.


Sometime in the last decade or two, whenever you signed up for your first cell phone contract, you became an untethered node in the electronic data network. Of course no one was talking about IoT then, but the trend – machine-to-machine communication and basic computation, often independent of human input – was already well underway. (What else but an early instantiation of IoT was the handshake between your phone and nearby cell towers that you never thought about unless you experienced a dropped call?)

Today cellular networks are more robust and almost no one mentions dropped calls. IoT talk, however, is everywhere. While consumer applications still get most of the media attention – wearables anyone? – there are now also lots of stories about nodes spreading to an amazing number of types of devices, equipment and buildings. Even to plants and animals. Connected cows aside, here’s my take on three changes associated with this transition that might present opportunity (and peril) for those of us in the semiconductor industry.

Cost. Surely this is the most significant issue related to designing, fabricating and deploying IoT devices. Consider the contrast to wireless communications and the collection and analysis of “big data,” particularly in the cloud, which continue to push the upper limits of big digital chips. Demand for these large SoCs created opportunities for chips that sold for relatively high prices. The IoT, on the other hand, suggests the need for much lower unit priced chips in very large volumes for the collection and receipt of information. These components will significantly challenge designers, especially given the new challenge of integrating analog, RF and digital logic into new types of packages with innovative MEMS sensors and actuators. Which leads to a second change…

Verification. Instead of verifying single chips, the emerging world of IoT physical verification means “system verification” for multiple chips, both analog and digital, communicating with one another in low cost, multi-chip packages, as well as communicating with other remote data collection sites. Some of the new verification tasks will be driven by a third big change…

Standards. Achieving high-volume products for the IoT requires agreement on standards that could lead to significant delays in full deployment. Cars are an important part of IoT but cannot readily notify each other of their location, velocity and direction until they are equipped with the appropriate systems built around industry standards. At the same time, there are a variety of standards to work with to improve the safety of cars and planes with driver-assist features that are taking off rapidly. While automotive manufacturers have always challenged the semiconductor industry for fewer defective parts per billion, these same automotive OEMs and Tier 1 suppliers are now demanding a new level of built-in self-test (BiST) for components, as well as test capabilities that can reliably monitor performance degradation over time, warn of impending problems, and send information back to the auto manufacturer. That level of quality requirement will be exceeded in some cases by the needs of the medical implant industry which is responsible for life-critical monitoring of pacemakers and insulin pumps and also for maintaining data security of the information transmitted.

It’s an exciting if fraught time in the industry. Sometime soon, the next generation of high-volume semiconductor components will emerge, eclipsing the volumes we see today for PCs, tablets and wireless handsets. This shift brings with it lots of opportunities. Among them: the need for new levels of ultra-low power dissipation (and energy collection by scavenging or other means), ultra-high reliability, integrated communications, integrated optical and MEMS-based sensors and massive data collection and analysis capabilities.

Despite the challenges, let’s hope the components that are beginning to appear will make life better, safer, cleaner and more convenient. And regardless, the end of dropped calls was just the beginning.

  • Jayna Sheats

    Excellent post; I only quibble with the assertion that “no one was talking about IoT [in the last decade or two]. Kevin Ashton says he first used the term in 1999, and Joel Birnbaum (VP of R&D at HP) was pushing what he called the Information Utility and Appliance Computing (which is exactly the same concept) in the early 1990s.

    It would be interesting to have a sociologist’s perspective on why it has taken so long. Obviously the now-ubiquitous smartphone had to develop according to consumer time scales, but why didn’t the Industrial IoT develop more quickly, and why did companies not start pushing connected “things” as part of the mobile phone value until the phone was already mature? Cost of the peripheral (sensors etc.) devices and their ease of use are surely critical.