The lightning growth of cloud computing may have begun to wane, but for its proponents the end of the ‘decade of cloud’ has a silver lining. New research by Stanford economist Nicholas Bloom shows that while the uptake of cloud by new firms has slowed, companies are ramping up their spending on the technology. This corresponds to a classic theory of the diffusion of innovation, in which a new technology first spreads across firms or consumers before deepening in its use.
Cloud computing as a concept was born in the mid-2000s, with the first recorded mention of “cloud computing” arising apparently in 2006 (see Regalado, “Who coined cloud computing?”, MIT Technology Review, 2011). Since that time cloud computing has seen an incredible rate of adoption across the US economy. Unlike many previous technologies such as the PC and the internet, the early adopters of cloud have been firms that are both small and large, young and old, earning cloud a moniker as a “democratic technology” (see my 2018 TNIT white paper, coauthored with Nico Pierri, “Cloud computing in the US: Democratizing Innovation”).
Chasing the Cloud
To evaluate the spread of cloud, I decided to look at three sources. First, adoption rates in US firms, as shown in Figure 1. The dataset for this analysis comes from Aberdeen Information’s call center, which has been making annual phone calls to millions of firms across the US since the 1980s. They painstakingly record the hardware and software used by millions of firms per year. This dataset is often used by academic researchers because of its broad coverage and high quality1. The records from more than 150,000 US firms clearly show a massive uptake in cloud computing since 2010, but also an apparent flattening off in growth of uptake since about 2016. Figure 2 shows this uptake by broad industry category and we can see a similar pattern.
Second, we analyzed mentions of cloud in the media, using Access World News which contains around 2,500 daily US newspapers going back to 2000. We could not directly search for the frequency of the word “cloud”, since this ended up tagging weather articles and uses like “His change in opinions clouded my view of the future…”.
So our approach, following on from Hasan et al. (‘Firm-Level Political Risk: Measurement and Effects’, 2019), was to search for two-word combinations as these are typically much more exclusive in meaning and therefore less likely to generate false positives. After extensive analysis we focused on around 30 cloud-computing bigrams, with the 15 most popular ones shown in Figure 3. These included combinations like “cloud based”, “cloud computing”, “cloud services”, “software as a service”, “cloud services” and “cloud solutions”.
Figure 4 pulls the frequency of cloud-computing bigrams in the press into a time series, showing a striking trend in terms of the percentage of articles mentioning a least one cloud bigram, rising from zero before 2006 up to about 0.6% by 2016. This striking rise shows how commonly cloud computing now seems to be discussed in the media, appearing as a term in about 1 in every 200 media articles. However, it also highlights how the growth in media coverage of cloud computing appears to be flattening off. According to media coverage, cloud was a new technology of the late 2000s and early 2010s. So by 2016 cloud computing had arrived - much like the cell-phone, internet or the PC - and it now appears to be viewed as a mainstream, widespread technology.
Finally, we also scraped the complete set of all company earnings call reports - the approximately 90-minute earnings calls that companies take with investors every quarter when they release their latest accounting earnings data. This text is useful for understanding what topics come up in a semi-structured business conversation. Conversations are defined as “semi-structured” because while the companies clearly are trained in advance of these calls and have rehearsed responses to many questions, the analysts are free to ask whatever they like and calls can sometimes veer off into unexpected topics. In Figure 5 we show the frequency of earnings calls that mention any of our bigrams used in Figures 2 and 3 (words like “cloud computing” and “software as a service”). We can see a rapid rise in business discussions of cloud computing from around 2008 to 2013, the period of the “cloud explosion” when this highly unusual new technology quickly became commonplace. As before, the earnings calls suggest that by 2016 the adoption rate of cloud technology began to slow down.
The Curse of Success
To sum up, cloud computing went from being one of the hottest new technologies around 2006 to widespread use and slowing rates of adoption by 2016. We should call this the “decade of cloud”. Does this mean the growth of cloud technology is over? No, far from it. Although the rates of new adoption of cloud appear to be slowing down, there is still rapid deepening of the use of the technology. (Clearly, not all cloud usage is the same. For example, firms could use cloud as a backup to cover 10% of their storage and compute costs, or as their primary application with on-premise compute as the 10% backup.) Gartner figures show exactly this is happening: Table 1 below reports cloud expenditure as a share of total IT expenditure, revealing that cloud is quickly becoming a central technology.
.jpg.aspx?width=350&height=183 "Table 1: Cloud expenditure")
As a result, total expenditure on cloud is accelerating, as shown in Table 2. This S-shaped pattern should be expected for any successful technology. The “S-shaped diffusion curve” was first popularized by Everett Rogers in his 1962 book Diffusion of Innovation and shows how growth rates typically start very low with initial enthusiastic adopters, pick up rapidly as the technology spreads to the mass market, then slow down towards the end as only late adopters are left converting. Cloud, like any other new technology, has been moving through the S-curve in its first decade, and appears to now be reaching sufficiently high levels of market saturation that future adoption rates are flattening off, with growth now coming from firms deepening their use.
This is in many ways the curse of success. Cloud has become so seductive and popular - and its uptake so rapid - that the industry is now running out of new customers, so the focus appears to be on the improvement of the technology and a deepening of its use.
Find Out More
Research by Nicholas Bloom on a range of topics such as innovation, management and IT, including his 2018 paper ‘Innovation, Reallocation, and Growth’, is available to view at:
https://nbloom.people.stanford.edu/research
The preceding is republished on TAP with permission by its author, Professor Nicholas Bloom, and by the Toulouse Network for Information Technology (TNIT). “Can Cloud Computing Keep Growing?” was originally published in TNIT’s June 2019 newsletter.
Footnotes
1) This database, formerly known as Harte Hanks (or Computer Intelligence InfoCorp) has been used in small samples in a number of prior papers – for example, Bresnahan, Brynjolfsson and Hitt (2002) and Bloom, Draca and Van Reenen (2015).
2) Gartner Source for Table 1: https://www.gartner.com/en/newsroom/press-releases/2018-09-18-gartner-says-28-percent-of-spending-in-key-IT-segments-will-shift-to-the-cloud-by-2022
3) Gartner Source for Table 2: https://www.gartner.com/en/newsroom/press-releases/2018-09-12-gartner-forecasts-worldwide-public-cloud-revenue-to-grow-17-percent-in-2019