Notes from the AI frontier: Applications and value of deep learning
Please create a profile to print or download this article.
Private Equity & Principal Investors
Avoid the temptation to select a platform simply because it has a particularly interesting initial use case. This would be like choosing a game console because it included a cool game in the box. Included applications matter but are only part of one element of a platform strategy. We have identified the top five characteristics of IoT platforms on which to base an evaluation. While these five are not an exhaustive list, they are the areas most likely to differentiate platforms in an important and sustainable way.
Sustainability & Resource Productivity
Eight shifts that will take your strategy into high gear
Businesses and developers have a bewildering array of platform options to choose from, which may have very different capabilities. The term platform is overused to the point where it doesnt convey much information beyond more assembly required.
is a senior adviser in the Silicon Valley office.
There are IoT platforms of every shape and size. There are platforms for specific industries like commercial real estate and family health. Some focus on one type of device: for example, there are at least two platforms focused on augmented-reality headsets. Some are focused on a particular function, like manufacturing. There is an IoT platform for dogs.
An executives guide to the Internet of Things
Meeting Societys Expectations
is a senior partner in McKinseys Montral office, and
Most broadly, a platform is software and hardware, which may include an operating environment, storage, computing power, security, development tools, and many other common functions. Platforms are designed to support many smaller application programs that actually solve business problems.
It is one thing to have a platform that takes data from your things and pipes it all up to the cloud for analysis by humans. Its another thing to run the analytics at the edge. Sometimes, the communications overhead of moving data to the cloud is onerous; transmitting terabytes of data from a remote mine or a ship at sea to the cloud could be prohibitive. Some platforms have specialized capability in handling this. Sometimes local autonomy is needed; some platforms allow you to take the human out of the loop and allow the platform to autonomously change the behavior of the connected endpoints or shift data only at convenient times. Moving applications from the cloud to the edge, and potentially allowing them to adjust operating variables like fuel flow or direction or temperature, may be a requirement.
Select topics and stay current with our latest insights
A new survey of global manufacturing experts explores the progress companies have made in implementing Industry 4.0, the challenges many still face, and the ways more players can unlock value.
Our team of 2000+ innovators bring new ideas, solutions, and services to our clients.
The IoT platform space is important, but crowded and confusing. How do you go about finding one that is right for your business?
In the meantime, choosing a platform should start with a good understanding of your IoT strategy. Identify the kinds of problems you are trying to solve, get a short list of likely solutions and use cases, and try to determine where you will need specialization and depth. If you have an idea of what kind of business problem you are solving and where the biggest challenges are, youll be able to quickly come to a short list of platforms (Exhibit 2).
The rate of adoption is accelerating. Here are six things you need to know.
Stay current on your favorite topics
In the Internet of Things, platforms are designed to deploy applications that monitor, manage, and control connected devices (Exhibit 1). IoT platforms must handle problems like connecting and extracting data from a potentially vast number and variety of endpoints, which are sometimes in inconvenient locations with spotty connectivity.
You may be content to have your data stored in the public cloud anywhere in the world with standard encryption. Or, it may be that for security or regulatory reasons, your data must be on your premises. Perhaps your data can be in the public cloud but only within certain political boundaries. You may have specific security requirements, either in the cloud or on your remote devices. There may be certain kinds of encryption, access management, or authentication that are required. Blockchain support may or may not be required. IoT platform capabilities vary here. Some are distinctive in certain areas of security.
Why so many platforms? Look at successful software platforms like Windows for operating systems. Platforms make a lot of money and are high-margin franchises that endure for decades. People and companies dont switch platforms very often. Often, switching costs are significant and platform choices persist for many years.
Our learning programs help organizations accelerate growth by unlocking their peoples potential.
Monetizing the flood of information generated by the Internet of Things requires a well-executed strategy that creates value.
Private Equity & Principal Investors
How to mess up your agile transformation in seven easy (mis)steps
Often, 80 percent of a data scientists time is spent combining, formatting, cleaning, and processing data to get it ready for analysis. Other companies have created new roles for data engineers, whose main job is to curate and cultivate data sources. Some platforms contain shortcuts or special tools that allow you to build a robust model of your important data much faster, reducing people costs and time to market significantly. Indeed there are some highly regarded platform companies that specialize in just this capability and use off-the-shelf technology for the other parts of the platform. Apart from the ability to conceptualize the data and understand what it is, also important is the ability of a platform to handle and manage a large number of high-velocity data streams coming from multiple different sources. The ability to handle vast, fast data may be critical, and there are some specialized technologies that focus only on that. Some are being licensed into different platforms.
Today, there is no one-size-fits-all best platform for every application. It may be years before the market anoints the winners in the IoT platform derby.
However, you may need to develop sophisticated IoT apps yourself. Platform providers dont understand your business problems the same way you do. Confirm that the application-development environment included in the platform is compatible with your own developers, or your trusted development partner. Make sure the development environment supports a way to containerize applications using a common service so that they can be ported to another platform should you decide to switch. Finally, you may need your platform to interface with large enterprise applications, like common customer-relationship-management or enterprise-resource-planning suites. Some platforms may include connectivity to popular CRM or ERP suites, and this may be an important feature depending on your IoT use cases.
As a result, many start-ups aspire to become platforms, because the winners create enormous shareholder value. Their investors push them to market themselves as platforms because winning platform companies can create 100-fold returns.
Sustainability & Resource Productivity
Discussion Paper – McKinsey Global Institute
Platforms are helpful because they abstract a lot of common functions away from the specific application logic. For example, regardless of whether you are trying to write an application to optimize fuel consumption or classroom space, a lot of the underlying technology needs are essentially the same. Application developers just want to focus on the specific problem they are solving and use common capabilities for computing power or storage or security. A good platform dramatically reduces the cost of developing and maintaining applications.
McKinsey uses cookies to improve site functionality, provide you with a better browsing experience, and to enable our partners to advertise to you. Detailed information on the use of cookies on this Site, and how you can decline them, is provided in ourcookie policy. By using this Site or clicking on OK, you consent to the use of cookies.
To get value from IoT across multiple use cases, it helps to use one (and only one) platform in your organization. The IoT platform market is immature and there are over 150 options to choose from. As this market consolidates, try to find a partner who is either large and will be in it for the long run or highly focused, distinctive, and successful in solving your most difficult problems. Look at the whole technology environment, not just the applications. Your most important requirement may be data wrangling, security, or local automation. Use fungible/off-the-shelf technology for the things that are less critical.
An analysis of more than 400 use cases across 19 industries and nine business functions highlights the broad use and significant economic potential of advanced AI techniques.
Getting the most out of Industry 4.0
Our flagship business publication has been defining and informing the senior-management agenda since 1964.
Choosing a platform is an important decision, because whether it is game consoles, smartphones, or the Internet of Things, its likely that whatever platform you choose will be with you for a long time.
Eight shifts that will take your strategy into high gear
There are three main application considerations when choosing a platform: what applications are available out of the box, what is the application-development environment like, and what are the common enterprise-application interfaces. Many platforms will include one or more applications that may be of some value out of the box, like the stock market or weather apps that ship with iPhones. Sometimes, very simple applications are the most popular. One manufacturing executive once told us, Id be thrilled to have an app that just told me what machines were on my factory floor and if they are switched on or off.
The second problem is that there can only be a small handful of winners in each platform space. Application developers dont want to learn multiple platforms. Businesses and consumers dont want to use and pay for multiple platforms. If there are 100 IoT platforms, then there is no platform, just aspirants. The market, over time, decides who the winners are, and the providers consolidate around two or three leaders.
In order to get value from theInternet of Things (IoT), it helps to have a platform on which to create and manage applications, to run analytics, and to store and secure your data. Like an operating system for a laptop, a platform does a lot of things in the background that makes life easier and less expensive for developers, managers, and users.
Please sign in to print or download this article.
Create a profile to get full access to our articles and reports, including those by McKinsey Quarterly and the McKinsey Global Institute, and to subscribe to our newsletters and email alerts.
Big IoT platform providers tend to also offer their own cloud hardware infrastructure (including storage, compute, networking, and data centers). For example, Amazon and Microsoft both provide a software-platform layer with IoT services, as well as a hardware-infrastructure layer that is broadly applicable across public cloud applications. The hardware infrastructure layer is capital intensive, has high fixed costs and significant economies of scale, and tends toward commoditization over time. As a result, most smaller platform players avoid offering it, providing only the software layer. They certify their platform on one or more of the leading public cloud providers. Many of the nascent platform companies may not be certified on all the major cloud providers (and often may run on only one of them). This is relevant for enterprises that may be seeking to standardize on a particular public cloud solution for other reasons. Make sure your IoT platform provider and your broader enterprise cloud strategy are compatible.
Creating a successful Internet of Things data marketplace
There are two main problems with this strategy. First, platform companies arent as focused on direct customer business value as application companies. A pure-play platform alone wont solve a business problem; an application is still needed. The platforms value proposition is harder to explain to business leaders. This translates into a higher cost of sales.
Our mission is to help leaders in multiple sectors develop a deeper understanding of the global economy.
Discussion Paper – McKinsey Global Institute
In many mature markets, there are often two dominant platform choices and a long tail of smaller players; for example, iOS and Android in mobile, Windows and Mac OS in desktop operating systems, and PlayStation and Xbox in gaming. But not in IoT, not yet. In IoT, sometimes it seems like there may be more platforms than things. Search Crunchbase for venture-funded IoT platforms, and you will get well over 100 hits. And that list doesnt include many bigger technology players entering the market with IoT platforms like Microsoft, IBM, and SAP or several industrial companies with similar aspirations like GE, Bosch, and Siemens.
Notes from the AI frontier: Applications and value of deep learning
