The industrial internet of things: Machine learning

economist.com

Manufacturers must learn to behave more like tech firms

THE best manufacturers used to be the firms that made the best widgets. No longer. As the “internet of things” spreads to the factory floor, products are being packed with ever more sensors and connected to the internet. That is transforming manufacturing—and the mindset that firms need to succeed.

The first shift is from products to services. By one estimate the number of wirelessly connected products in existence (excluding smartphones or computers) will rise from 5 billion today to 21 billion by 2020. The data these products generate are the raw material for new services—from windscreen-wipers whose movement helps produce real-time weather reports, to tennis racquets whose sensors tell you why your backhand isn’t working, or machines that can order a new spare part when it is needed (see article). Such services will often be more profitable than the products they are based on.

The second, related change is the race to develop “platforms”, a software foundation upon which lots of services and applications can be built. In the world of information technology, platforms are already a well-known concept: think of the operating systems for smartphones, such as Apple’s iOS and Google’s Android, which host all manner of apps. In industry the idea of platforms is new, and menacing. If Apple or Google, for instance, were to control entertainment systems in cars, and the data they throw off, many carmakers would risk becoming the computer-makers of the road: churning out bits of metal while others grab the really valuable parts.

This is disorientating stuff. Manufacturers are used to a world in which they take materials from suppliers, turn them into products and push them out to customers. But for clues on how to embrace the industrial internet of things, Germany offers early lessons. The country is an industrial powerhouse—manufacturers account for 22% of GDP, compared with 12% in America. In 2011 Germany launched “Industrie 4.0”, a government initiative to promote the computerisation of manufacturing (see article). Several companies have launched software platforms: Trumpf, a maker of industrial equipment, has set up one that collects data from machines built by it and by rivals, in order to help customers ensure that their production processes run more smoothly. Bosch helps other firms create services based on internet-connected devices.

For many manufacturers—in Germany and beyond—the principal sticking-point in making this digital leap is often cultural. They have to forge a more open type of relationship with competitors; in August, for example, BMW, Audi and Daimler, three German carmakers, banded together to buy a digital-mapping company. Companies also need to become less hierarchical and more entrepreneurial; Klöckner, a metals trader, has set up an incubator for startups far from its headquarters, to give it more freedom. All this may be old hat for IT firms, but it is new ground for many manufacturers.

Platform views

Governments, too, need to adjust. Bureaucratic initiatives like Industrie 4.0 matter less than the basics: top-notch broadband infrastructure, the right balance between open data and privacy (see article), and decent computer-science teaching in schools, an area in which Germany falls short. More ambitious ideas might include “special digital zones”—reserving designated bits of cities to try out self-driving cars or commercial drones. There is no single recipe, but both bureaucrats and industrialists must grasp one fact: making things is not what it used to be.