Lots of vendors are eager to sell enterprises an \u201cIoT platform,\u201d but it\u2019s not always clear exactly what those \u201cplatforms\u201d actually do, why you need one, and which one you should choose. As Hackernoon put it in April 2018:\n\n"We\u2019re a cross-functional, fully integrated, full-stack, serverless, hardware agnostic, AI, IoT platform that offers you infinite infrastructure\u00a0.\u00a0.\u00a0.\u201c said every confusing IoT platform website\u00a0ever.\n\nMultiple overlapping IoT platform definitions\nSo, what is an Internet of Things (IoT) platform? Observers don\u2019t always agree \u2014 in fact, they don\u2019t always agree with themselves! For example, from that same Hackernoon post: \u201cAn IoT platform is an integrated service that offers you the things you need to bring physical objects online.\u201d\n\nBut in 2019, a newer Hackernoon post had this to say: \u201cThe term \u2018IoT Platform\u2019 is really too broad to be useful to most people.\u201d Worse, it cites the 2018 edition of Gartner\u2019s Hype Cycle for Emerging Technologies: IoT platforms have crested passed the \u201cPeak of Inflated Expectations\u201d and are ready to fall into the \u201cTrough of Disillusionment.\u201d\nIoT for All, meanwhile, says: \u201cIoT platforms are the support software that connects everything in an IoT system.\u201d In this model, IoT platforms:\n\nConnect hardware, such as sensors and devices\nHandle different hardware and software communication protocols\nProvide security and authentication for devices and users\nCollect, visualize, and analyze data the sensors and devices gather\nIntegrate all of the above with other web services\n\nMaybe Postscapes\u2019 definition is simpler: \u201cIoT data platforms offer a jumping-off point by combining many of the tools needed to manage a deployment from device management to data prediction and insights into one service.\u201d\nBut I actually think this 2016 description from Link Labs gets closer to the mark:\n\nAn Internet of Things (IoT) platform is the support software that connects edge hardware, access points, and data networks to other parts of the value chain (which are generally the end-user applications). IoT platforms typically handle ongoing management tasks and data visualization, which allow users to automate their environment. You can think of these platforms as the middleman between the data collected at the edge and the user-facing SaaS or mobile application.\n\nThat last line is key because to me, an IoT platform is little more than a fancy name for the middleware that connects everything together. i-Scoop focuses on that aspect: \u201cAn IoT platform is a form of middleware that sits between the layers of IoT devices and IoT gateways (and thus data) on one hand and applications, which it enables to build, on the other.\u201d\nPerhaps, though, IoT platform vendor KAA offers the most honest description. While acknowledging the middleware aspect, the vendor also allows that \u201can IoT platform can be wearing different hats depending on how you look at it.\u201d\nHundreds of IoT platform vendors\nStill, the confusion isn\u2019t stopping vendors ranging from Amazon Web Services (AWS) and IBM to Oracle and Salesforce from selling IoT platforms. In fact, back in 2017, IoT Analytics compared a whopping 450 IoT platforms. (That number was up from 260 in the firm\u2019s 2015 analysis and 360 in 2016. But though leading products are growing at more than 50 percent a year, the market remains highly fractured.)\nThe bigger question, though, is exactly what is IoT Analytics talking about when it looks at IoT platforms? The firm separates IoT platforms into five categories:\n\nIoT Application Enablement Platforms\nIoT Device Management Platforms\nIoT Cloud Storage Platforms (IaaS)\nIoT Analytics Platforms\nIoT Connectivity Backend (Platforms)\n\nMakes sense, right? But in the most recent Hackernoon post, there are just two kinds of IoT latforms:\n\nBroad-based IoT platforms from cloud service providers like AWS, Microsoft Azure and many others.\nSpecialized IoT platforms built on top of the cloud service providers, including GE Predix, Siemens Mindsphere, C3 IoT , SAP Leonardo IoT, Cisco Kinetic and many more.\n\nWell, that description makes sense, too \u2014 as far as it goes. But wait, there\u2019s more! The 2018 Hackernoon post cites four kinds of IoT platforms:\n\nEnd-to-end IoT platforms\nConnectivity management platforms\nIoT cloud platforms\nData platforms\n\nEven more confusing, Hackernoon contend that those platforms can be separated into four vertical markets:\n\nHobbyists\nConsumer electronics\nIndustrial IoT solutions\nIndustry-driven vertical market solutions\n\nWhy you (may) need an IoT platform\nSheesh, this is getting really complicated. So, let\u2019s try to make it radically simpler.\nThe Internet of Things Wiki recently updated its list of top IoT platforms\u00a0and used the post to explain that in an IoT implementation, \u201cThe gap between the device sensors and data networks is filled by an IoT Platform.\u201d The point of such a system, obviously, is to help enterprises fill that gap more quickly, effectively, and efficiently than they could by building those solutions for themselves.\nBut because the category is still evolving, Hackernoon advises enterprises to treat \u201cemerging technologies, such as IoT platforms, \u2026 more like strategic partnerships than assets or commodities.\u201d Makes sense to me. After all, can something you can\u2019t even clearly define really be a commodity?