The Rise of the Internet of Goods: A New Perspective on the Digital Future for Manufacturers >> HOME

Digital Versus Physical Industries


Entrepreneur and venture capitalist Marc Andreessen famously said in 2011, “software is eating the world.” So far, he is half right. Software has devoured any industry where the final output can be easily reduced to bits. These are the digital industries—including communications, entertainment, finance, and even professional services. The full content of a daily newspaper can be put into a small digital file.

But so far software has not been able to eat the physical world. Data is important for physical industries like manufacturing, construction, agriculture, and healthcare, but it is not the main story. The construction of a building requires huge cranes, not just a digital twin of a crane. A physician treating a patient needs an actual tool like a surgical knife, a laser, or the appropriate drug. And a company building an airplane needs to work with materials that won’t fall apart in flight.  

This divergence between the digital and physical sector has several important consequences. For one, companies in the digital sector invest far more in software and information technology equipment. In 2016, 65% of U.S. software investment went to the digital industries, which make up only 35% of private sector GDP and 30% of private sector employment.*

The digital sector is also outperforming the physical sector on a wide range of economic measures, including faster productivity growth, faster job growth, and faster wage growth. For example, between 2000 and 2017, productivity growth in the digital sector averaged 2.5% annually, compared to only 0.7% in the physical sector. 

The obvious question: why haven’t physical industries embraced digitization more enthusiastically? Making the business case for deep digitization has proven to be tough in many industries. Take healthcare, for example. Lockheed developed the first electronic health records (EHRs) in the 1960s. But it was hard to make a business case for the EHR, and widespread adoption did not take hold until spurred by the federal government. Even now, it’s not clear if the current version of EHRs contain all the clinical information that could be used to track treatment outcomes.

Similarly, most trucking companies have adopted GPS to keep track of their vehicles. But these relatively minor investments have not fundamentally changed the business model of the short-haul or long-haul trucking industry. Autonomous trucking is transformative, but widespread use of fully driverless trucks is further off than people expect.

There are two good examples of the successful application of data to transform a physical industry. The first example is oil and gas mining. The ability to use data to visualize and analyze oil and gas formations made horizontal drilling cost-effective and opened up vast new reserves that were not accessible before. In the U.S., proved oil reserves went from roughly 21 billion barrels in 2006 to 33 billion barrels in 2016, according to the Energy Information Administration. Natural gas reserves grew at roughly the same rate, from 220 trillion cubic feet in 2006 to 341 trillion in 2016. In effect, data-driven oil and gas exploration and extraction increased the size of oil and gas reserves in the United States by roughly 50%.

Retail, or more precisely the distribution of goods to the household, is the other physical industry transforming by data. Historically it has always been too expensive to deliver most goods directly to households, so the solution was to have individuals pick up their goods at central storage locations—that is, stores.

In the 1980s and 1990s, big box retailers such as Walmart and Costco took that trend to the natural conclusions by effectively turning their stores into large warehouses, which reduced costs and let consumers do their own “picking and packing.” In that era, retailers developed sophisticated back-office digital solutions for managing their supply chains. But store employees were mostly doing the same tasks as they had always done—re-shelving inventory and ringing up the register. The result was an increase in low-wage workers.

The first wave of ecommerce pioneers in the 1990s used technology to improve the ordering stage of retail. The use of websites made it possible to order goods such as groceries (Webvan) and pet supplies ( online, which seemed like a great innovation at the time.   

However, it turned out that the main point of retail was to get goods into the hands of consumers, which was more difficult and expensive than it seemed. Webvan burnt through $1.2 billion in capital before going bankrupt in 2001. The company had a vision of robot-enabled fulfillment centers but did not realize how hard the task was.
Building on the lessons learned by Webvan and others, Amazon’s great innovation was to apply robotics and machine learning to greatly improve the productivity of order fulfillment, picking and packing of individual items, and the best allocation of items across fulfillment centers.  

It’s worth noting that Kiva, the warehouse robot company bought by Amazon in 2012, was founded by a former Webvan manager, Mick Mountz. Mountz recounts his experience in talking to venture capitalists.

“Still, in 2003 even I knew why VCs weren’t biting. They looked at Kiva and saw a company that was complicated, because it was both a software and a hardware business. Software they liked. In fact, I remember potential investors saying to me, ‘You’ve got so much data at the heart of this thing. Can’t we just sell the data?’ (We couldn’t. We needed to physically move inventory around the warehouse to create value.)”

Amazon’s technological advances in cutting costs and boosting productivity were then transformed into new business models. For one, the deceptively simple promise of two-day delivery at no extra cost for Amazon Prime members made e-commerce shopping almost as fast as bricks-and-mortar, in terms of consumer perceptions. Fulfillment By Amazon (FBA) allowed Amazon to extend the same promise to goods sold by other online merchants.

Suddenly, e-commerce became a much better proposition for consumers. Rather than driving to the mall, parking, walking through the store and looking for the right aisle, waiting to check out, and driving back home, a harried parent could simply go online and get the desired goods within two days. In effect, they could pay someone else to do their driving, picking, and packing for them, at a price that was appealing.

In fact, Amazon boosted productivity so much in the ecommerce fulfillment centers that it could offer Amazon Prime at a low enough price to attract 100 million subscribers worldwide. And as demand soared, fulfillment center employment soared as well. Many fulfillment centers employ thousands of people to work alongside the robots. As a result, Amazon became the fastest private company to employ 300,000 workers in history, and its job growth curve since its IPO in 1997 looked identical to the first 20 years of General Motors.

At this point, e-commerce offers the first full-scale example of how digitization can transform a physical industry. There was no magic wand of data that suddenly made everything different. Rather, years of hard work and incremental improvements increased the productivity of picking and packing enough to make new business models economically feasible.


*The digital industries, in this definition, include computer and electronics manufacturing; the entire information sector, including software, telecom, and Internet search and publishing; the finance and insurance sector; the professional and technical services industries; and management establishments.  Physical industries include the rest of the private sector, including manufacturing, construction, transportation, mining, and healthcare.