The Internet of Things (IoT) or as Cisco calls it, the Internet of Everything, has been the theme du jour for almost every semiconductor-related industry event in the past year, and going forward for 2015. The IoT hype continues, as many see the opportunity in this global trend to connect people, processes, data and things that will require 50 billion connected devices by 2020, and will generate a $19 trillion revenue stake over the next decade, according to Li Li, Technical Leader, Manufacturing Technology Group, Global Supply Chain Management, Cisco Systems. One of the big questions on my mind and others is, who exactly are these stakeholders? What does the chip industry stand to benefit from with the realization of the IoT?
At ECTC 2015, which took place last week in San Diego, May 27-31, Jan Vardaman, TechSearch International Inc, lead a plenary session that included Li; Ilyas Mohammed, Jawbone; Jerry Tzou, TSMC; and Subu Iyer, UCLA; all who provided insight from different sectors of the market including wearable devices, chip manufacturing, packaging R&D, and the networking infrastructure. The fifth panelist spot was dedicated to a sign labeled SECURITY in large block letters, indicating that this topic in reference to the IoT deserved its very own seat.
As ECTC attendees are mainly interested in the component and packaging aspects of the IoT, Vardaman posed questions like what are the challenges beyond or related to packaging? What are the limiting factors? What does this mean for the packaging community? Is system in package (SiP) important in this area? Who is going to be doing the SiP?
“I’m not sure the IoT will radically change the way we do things (in the packaging community,” said Iyer. “Rather, the IoT is a consequence of what we’re doing.” He added that it’s not just about packaging sensors; we need to think more about entire system requirements. The IoT/IoE integrates sensors with cognitive ability that needs reliable power and reasonable communication capability for data and information.
Tzou noted that system in package (SiP) would be the packaging trend, because with all the different technologies required, it can’t all be put in one device. At TSMC the IoT means not pursuing Moore’s Law, and instead focusing on integrating mature technology nodes such as image sensors and microcontrollers. InFO, 3D IC, and CoWoS technologies are all packaging integrators that will enable such system integration.
According to Mohammed, exactly who is going to do the SiP – the OSAT, EMS, IDMs – is a problem Jawbone struggles with. “The best thing would be a short loop from IC design to product,” he noted. They tend to select chips from different vendors, and it’s a 9-12 month effort to ship wafers. They are interested in finding someone who can do it very quickly, he said.
“The IoT is an infrastructure, and the data center is mission critical,” noted Tzou. The reliability requirements are stricter for the data center and automotive IoT than wearables that get changed every year or two. There’s not going to be “IoT-certified” packaging technology; it will all be very application driven — it could be wire bond, it could be flip chip.
And as it is of critical importance across the board, Vardaman also asked the panelists about security.
As Jawbone manufactures wearable devices that gather people’s data and transmit it to smartphone apps, Vardaman posed the question to Mohammad, “What about security for people’s data?” to which he responded. “We are not monetizing personal data. The data is there, but the only time it is visible is on the company’s blog post.” Li added that from Cisco’s perspective, they have experience in managing security from the first and second waves of the Internet, which were about connectivity and then web browsing, email, and search engines.
Iyer somewhat downplayed the security issue, acknowledging that the scale of things talking to each other will increase, and that these things will be unreliable. However, while data can get into the wrong hands, data is not information all by itself. “The IoT is not just about the things and the data, its about converting raw data into information that is contextual,” he said. “It’s about using that information to make decisions, possibly without human intervention.” This is what will make the IoT successful.
None of the panelists directly answered what we mean when we ask about security; the risk of making all our personal data available to whoever wants to find it. The news is already rife with stories of identity theft carried out by hackers; will this situation worsen as we come to depend more on using the Internet to transfer information?
The most pointed question came from the audience: Who’s going to make money on the IoT? If the devices consist of classic components that are sold for a few cents, it’s not likely to be the semiconductor device manufacturers. Although many say that there will be value add in functional integration via advanced packaging.
Li talked about a business model where a particular platform connected to the wearable device has an available service that can be sold.
Iyer suggested that like the cell phone market, the end product is information. “People who can deliver information and data; they will make some money,” he said. “Everybody will make a little money. Nobody will make a lot.”
During an interview after the panel, I asked Iyer what he thinks the chip industry stands to gain from the IoT. He says the chip industry will benefit because of the sheer volumes required to make the IoT happen. He predicts new life for legacy devices and chips, and lots of heterogeneous integration. Will 3D be involved? “Things don’t need 3D, but the data analytics will need it,” he said.
Whether we’re using our smartphones to transmit diagnostic information to our primary care physician, or checking our smart home setting remotely, we will need access to data. Currently we do this through our cell phone providers, or by accessing WiFi hotspots. But how many of us, myself included, switch our cell data roaming off to save money when traveling, or end up in remote places with sketchy WiFi service. Its one thing to be disconnected for a few days when we don’t rely on being connected to conduct our every day lives. What happens when we are so connected, that we truly can’t live without the Internet? Who will pay for the data? Is the WiFi infrastructure sufficient?
Iyer says this is not a big challenge. Whoever has the most at stake will pay for the data. He cited the Tesla model. Each Tesla vehicle comes with 3G connectivity for life. The company needed to offer navigation services, so built the cost into the price of the car. When you buy an interconnected home, you’re already paying for the data plan. He predicts that WiFi will be viewed as a utility.
Concerns notwithstanding, perhaps what’s most important to consider is that even if, as many point oht, that we’re just doing what we’ve been doing all along, the IoT can not be realized without the semiconductor and related industries. But consider that the real value in the IoT is not the revenue it will generate, but the societal benefits that can be gained from 50B connected devices. ~ F.v.T.