What’s going on in the cloud? That’s been a question for over a decade, not so much because we’ve not been able to answer it, as because the answer seems to be constantly changing. There is a lot of change in the cloud, as enterprises and even cloud providers tell me, but there’s also a lot of hype and misconception, and a lot of conflicting Wall Street and industry analysis. Can we answer the question with any confidence? We’ll see.
The major cloud providers (Amazon, Microsoft, and Google) don’t provide a lot of granularity in their reporting of cloud revenues. Enterprises tell me that their 2024 budget for cloud spending is up 9% over last year, and last year was up 13% over 2022. Actual cloud spending in 2023, they say, was up by 15%, which suggests that spending growth this year could be expected to be up by perhaps 11%, still a bit down y/y.
One reason for the overshoot is the “tactical” nature of the cloud, something I noted in an earlier blog. If the largest source of new spending in IT budgets are ad hoc, fast-turnaround, requirements set by senior management (recall that, as I said yesterday, 83% of current IT budgets just fund current applications, 11% fund ad hoc projects, and 6% fund true IT modernization), then it’s difficult to address this source through anything other than cloud hosting.
The discrepancy between cloud growth and my enterprise numbers likely comes from startups and OTT users of the cloud. This has always been the biggest chunk of revenue for Amazon, and it’s been declining as startup issues have grown last year and this year. On enterprise cloud spending, my own data has Microsoft on top for 2022 and 2023, Amazon second, and Google third.
The expected dip in cloud budgets for this year was planned, because as I’ve been reporting, enterprises have been increasingly anxious about cloud spending, and increasingly likely to actually “repatriate” applications previously moved from the data center to the cloud. This, however, has not targeted those extemporaneous applications done in the cloud for reasons of short turnaround. It appears that in roughly two-thirds of cases, these things were never in the data center at all, and they’re really new elements of existing data center applications added in the cloud because it was faster to do that way.
The nature of the primary force behind cloud growth, that extemporaneous stuff, means that the new things being done on the cloud are more likely to use web service APIs rather than only vanilla VM/IaaS or container hosting. Faster results are critical to the activity, apparently. There’s only limited data on whether there’s any drive to redo these new application components, but so far not much has developed. This might be an indication that once any model of hosting has been picked, there has to be a major impetus for it to grab a piece of that 6% IT-driven redo funds. If so, then this extemporaneous application drive may become the next major source of angst about cloud spending, since enterprises generally say that using these custom cloud API features is more expensive than building your own capabilities on top of IaaS or containers.
There are sharp differences in enterprise cloud views, and even usage patterns, depending on firm size. The great majority of enterprises are hybrid cloud users. Among large multi-national firms, multi-cloud dominates over single cloud, but for enterprises operating in a single market geography (meaning over 85% of their business is there), a single provider is the main choice. Interestingly, more enterprises say they are reducing their number of cloud providers than say they are increasing that number.
Mid-sized businesses tend to use the cloud more than enterprises, if they use anything other than personal computing, because they typically don’t have data centers per se. Those with any form of central computing often get vertical applications from a VAR, and these channels also provide the necessary hosting, either small servers or the cloud.
Among enterprises, the biggest cloud-related questions relate to the relationship between the cloud, the data center, and the VPN. Applications have, from the first, been viewed as input-process-output layers within transaction processing, and “reporting” or “analytics” as a background activity. Most enterprises still keep their mission-critical databases and the portion of transaction processing and background applications that relate to those databases in their data centers. Most have never tried to move them, nor do they expect to. What most also realize is that many applications have an “input” front-end element that may support considerable user/worker interaction without involving transactions to the databases. This is the stuff that they find most suitable for cloud hosting, and they’re now working harder to align their cloud plans (other than the extemporaneous stuff) with that model.
The question, obviously, is how extensive this cloud piece will become. That depends on how much of each enterprise application is susceptible to the cloud value proposition, and that in turn depends on how many employee-facing applications are cloudified rather than remaining in their original form. The amount of work that flows through the cloud, in both absolute and relative terms, sets up answering the question of relationships.
Suppose there was no data center because everything went into the cloud. There’s also no VPN, because everyone connects to applications via the Internet. Now suppose nothing is in the cloud. The data center and VPN are the only game in town. You can infer that the more work goes through the cloud, in particular the greater the percentage of all work that does, the greater the chance that the cloud becomes the enterprise network, period. It’s not necessary that the data center is eliminated, only that it’s the back end of transaction processing to the cloud’s front end.
Right now, only 11 of the enterprises I chatted with seem to be parsing this question, and nobody said anything about having resolved it. I’m not sure, given the enterprise loss of proactive IT planning initiatives, that anyone is now or is likely to be tasked with finding the answer, either. That doesn’t mean they won’t, only that the answer will become clear as cloud directions take their course.
