Hardware commoditization and the IoT service model

As hardware margins continue to shrink, system developers must explore new ways of monetizing their products.

Earlier this year, economist released the book “The Society: The , the Collaborative Commons, and the Eclipse of Capitalism.” In it, Rifkin argues that the Internet of Things (), which he defines as a unison of the Communications, Logistics, and Energy Internets, will converge with the competitive capitalist market to usher in a period of extreme economic productivity in which “the of actually producing each additional unit – if fixed costs are not counted – becomes essentially zero, making the product nearly free.” As a result, capitalism as we know it today will be slowly replaced by the distributive economic model of the Collaborative Commons.

While this notion may be objectionable to those of you in the Western world, there’s no denying that the cost of compute and are in a sustained decline. Moore’s Law continues (at least for now) to eat away at the margins of vendors, and is currently providing free 5 Mbps Internet in Austin, Texas, Kansas City, Missouri, and Provo, Utah, with 1 Gbps speeds available for $70 per month. Trends like these have led to a lot of business model rethinks in the tech sector, with many companies turning to the for answers.

The cloud space has become a crowded one to say the least over the past couple of years, partially because of the “services” model it offers businesses. Today cloud service models range from Software-as-a-Service to Platform-as-a-Service to Infrastructure-as-a-Service (, , and , respectively), with the newly coined Everything-as-a-Service () entering the fold as well. These service platforms deliver everything from storage and security to full-blown end-user applications, which can each be neatly packaged as line items on a monthly statement.

So why is the cloud important for embedded developers? Hardware commoditization.

Hardware commoditization and the

As the dust settles around IoT standardization, open, modular architectures with an emphasis on software development and app enablement will take precedence over custom or application-specific hardware designs (look at the success of “maker” boards like the Raspberry Pi). Does your next system require connectivity? Order a module from Shanghai. Do you also need sensors? Browse the capes on ’s website. If Rifkin’s predictions hold true, specialized hardware will only be sustainable in a very narrow set of fringe applications, so the majority of system developers will have to find other ways to create value.

Take, for example, ConnectSense, a company based out of Naperville, Illinois that produces a line of Wi-Fi sensors for home and building automation (www.connectsense.com). ConnectSense sensors range from temperature and humidity to water, motion, light, and dry contacts, but the target market demanded a cost-conscious approach across the product line. Therefore, the company organized the portfolio around a base platform consisting of a repurposed ARM7 that was developed in-house, a TI MSP430 MCU, and a low-cost, low-power Wi-Fi module from partner Shanghai High-Flying Electronics Technology Co., Ltd. This approach allows multiple sensors to be manufactured quickly and easily with only few modifications to the common platform.

What makes an architecture like ConnectSense unique, however, is that it’s also powered by a proprietary cloud platform that handles most of the heavy lifting of software and application development, so additional hardware resources aren’t required on the physical sensors themselves. For novice users, the ConnectSense cloud provides an if/then rules engine that can be used to set up alerts via email, text message, phone call, webhook, or tweet in a plug-and-play fashion, while more advanced developers can take advantage of a full REST API (Figure 1). Today the ConnectSense system is being leveraged in applications such as datacenter monitoring and agricultural observation.

 Figure 1: The ConnectSense cloud allows users to use a rules engine or full REST API to monitor readings such as temperature and humidity over time.
(Click graphic to zoom by 1.9x)

ConnectSense sensors currently start at around $150 a piece, with the cloud platform included at no extra cost. But what’s really intriguing about this architecture is that if hardware margins erode substantially enough, the company could conceivably flip the current model on its head and monetize their cloud services while providing ConnectSense sensors free of charge (sort of like your cable provider). I guess you could call this “IoT-as-a-Service.”