What may be next for Cloud (IaaS) growth?
$ORCL $MDB $ESTC
TLDR: Recent news about slowing growth by major cloud IaaS (Infrastructure as a Service) vendors has sparked questions on what is next for public cloud IaaS growth. The Cloud IaaS has shown tremendous upside, growing at ~25% CAGR between 2012-2021. But like any new technology it may be ready to exit its hype phase and enter a period of rational expectations. This article makes the point that cloud growth will continue but at slower rates over the next 3-5 years for the following reasons: increased barriers for cloud workloads to show ROI value, more competition from new players leading to drops in pricing and thus lower margins, on-premise workloads proving stickier than common belief - and, finally, a new paradigm - Edge computing where sensors and edge devices (think iPhone) do some of the computing, reducing the need to move to cloud. Thus, investors who want to invest in cloud IaaS growth can consider hedging their bets by investing in companies which have healthy businesses in both cloud IaaS as well as on-premise environments.
Cloud computing IaaS has shown tremendous upside over the last ten years, growing revenue from $6B in 2012 to $90B in 2021. The question is where it goes next. Recently major enterprise cloud IaaS vendors have warned of slowing growth with growth rates guiding to slower growth (10-15% vs. more than 30% in 2020), The big question is whether this slower growth is here to stay or a minor blip with faster growth (>30%) soon returning.
Cloud (IaaS) growth will continue, but at a more moderate pace
Firstly, let us acknowledge that Cloud computing IaaS will continue to grow for the coming years due to the benefits it provides; public cloud IaaS supports enterprises’ ability to flex computing capacity up and down more efficiently to deal with spikes/growth in computing workloads (e.g., for an online game that gained popularity), thus enabling enterprises to focus more on their product and its iterations. However, this article makes the case that public cloud computing growth will moderate and the recent guidance of slower growth may persist for the next 3-5 years.
In this article we mainly focus on prospects for IaaS, but let’s differentiate it from other cloud services. IaaS (Infrastructure as a Service) is mainly about compute, storage and networking and replaces the old school data centers (e.g., Amazon Web Services). The second important service - PaaS (Platform as Service) - is about enabling developers to build applications on top of IaaS. Finally, we have SaaS (Software as a Service), where vendors deliver an end-user application - applications like Okta, Salesforce and Workday. An important point to note is that hardware has historically had lower margins vs software, and IaaS best represents the substitute for data center hardware, while SaaS is the best representative for license software.
Insight #1: Cloud IaaS has graduated from its hype phase
Every technology goes through its hype phase.The Gartner Hype cycle, as shown below, indicates technologies rise to a peak of inflated expectations, fall to a trough of disillusionment, and then finally ascend a slope of enlightenment before reaching a plateau of productivity. As we present with examples below, companies are no longer taking cloud decisions as a given but instead evaluating each decision with more rationality. One can argue that with companies now increasingly evaluating the ROI of their spend, Cloud IaaS has passed its hype phase, and is entering the phase of more realistic expectations.
First, some, like Basecamp CTO, David Hansson argue it is expensive: ‘renting computers is (mostly) a bad deal for medium-sized companies like ours with stable growth’. Entrepreneur turned Venture Capitalist Martin Casado (May 2021) argues ‘cloud clearly delivers on its promise early on in a company’s journey, the pressure it puts on margins can start to outweigh the benefits, as a company scales and growth slows’. He gives examples of enterprise software vendors like Dropbox who have repatriated workloads from the public cloud completely and some like Crowdstrike and Zscaler who have taken a hybrid approach of running workloads in both their own data centers and the public cloud.
The other side of the argument asserts that cloud ROI problems are self-inflicted and mainly due to companies not adopting the right cloud architectures and not having the right cloud cost controls in place. In a 2022 Andot report on cloud costs, 53% of respondents stated their biggest challenge to cost control is in gaining true visibility into their cloud usage. Additionally, finding the right cloud architecture for each workload category is challenging. Thus, companies’ internal operational challenges may also contribute to cloud IaaS’s moderating growth.
Finally, companies may have to decide which workloads are best suited for the cloud, creating decision points that may slow growth. David Linthicum, an industry thought leader, talks about 2023 being the year of cloud repatriation. He makes the case that ‘It makes the most sense to repatriate workloads .. that typically do a lot of the same thing, such as just storing data for long periods of time without any special data processing (e.g., no advanced artificial intelligence’.).’
The forthcoming economic soft period will force people to make rational choices on where the cloud helps and where it does not.
Insight #2: Though the public cloud has been growing faster, the real surprise has been that the data center hardware and on-premise software continue to grow in tandem with the public cloud.
The conventional wisdom in investment media is that public cloud IaaS is going to take the market share from data centers and people are going to ultimately move all workloads from on-premise to the cloud. However reality for the past few years has been a bit more nuanced.
Yes, there is momentum in the cloud but on-premise workloads continue to grow. Leading companies continue to match their workloads to best-fit platforms rather than migrate exclusively to the cloud."According to Synergy Research, data center spending grew 3% in 2021. This growth is not a complete surprise.
Here is Jamie Dimon, the CEO of the biggest bank in the USA, in Jan 2022 - “So we spent $2 billion on brand new data centers, okay, which have all the cloud capability you can have in private data centers and stuff like that. We’re still running the old data centers”. Thus not all companies need to move to public cloud, the big ones can build the right architecture themselves to take advantage of cloud computing benefits.
One can take a deeper look at financials of Elastic and MongoDB, which both have robust data businesses on-premise as well as in public clouds. Both these vendors started as on-premise vendors but are now aiming to be cloud-first businesses. Elastic’s non-cloud aka self-managed revenues grew 14% YoY (as of Elastic’s Q2 FY23) and their non-cloud revenues are still a majority of their revenues (56%). Similarly, MongoDB’s non-cloud subscriptions revenue increased 26% in the nine months ended in Oct 2022. It is important to note MongoDB’s cloud revenues exceeds their on-premise software revenue, still, in their recent earnings call (March 2023), Michael Gordon, MongoDB CFO acknowledged the growth in their on-premise revenue (EA) - ‘if you think about over the course of the full year, EA really is what drove a lot of the outperformance that we've seen this past year in fiscal 2023’.
One can make an assumption that companies are making rational decisions on whether it is more prudent to move to the cloud or stick with the status quo. For verticals like healthcare and financial services, where the rate of change in business processes is a lot slower than gaming or social media, staying in existing data centers may be a rational choice based on their circumstances.
Finally if we take a look at history, mainframe computers are still running strong after 70 years. Some articles indicate that 67 out of Fortune 100 enterprises (banks, insurers and airlines) still use mainframe and other heavy duty applications. Thus, borrowing a history lesson from mainframes, one can argue that enterprise software workloads are sticky and thus on-prem self-managed workloads may have more staying power than expected.
Insight #3: History suggests Iaas pricing and margins are probably going to decline
We start with another look back at history. Flash back to the 1990s and 2000's where hardware players like EMC, Sun, and HP either got gobbled up or became a sliver of their glorious pasts. The reason - while hardware competitive advantages are competed away over time, software advantages are more permanent. Comparing the hardware players to the software players from the early 2000s, Oracle, SAP and Microsoft still have healthy businesses.
Moreover, hardware margins have historically been lower than software margins. Why does it matter to IaaS? Most of IaaS revneue is constituted by compute and storage revenue, analogous to what older Server players EMC, Sun and HP provided. If we think the past will repeat itself, public cloud vendors' margins will decline. We cannot predict how much the margins will decline, but for HP Enterprise’s compute revenue operating margins are 17%.On the other hand, public cloud IaaS operating margins have ranged ~ 25% and ~30% for the last year, but margins have been on a downward trend.
Recent announcements from Uber and the Department of Defense have new entrants Google and Oracle playing equal fitting vs. the two incumbents. Thus, IaaS does not seem to be a winner take all market; instead, more competition will likely lead to pricing pressure and decline in margins. For example, Google and Oracle have other profitable businesses and thus have incentive to take advantage of the recent economic environment where costs are important to steal market share based on price.
If history is any guide, more competition will drive down prices and negatively impact revenue growth for Cloud IaaS revenues. As mentioned earlier, it is important to reiterate that this article’s view is that Cloud IaaS will continue to grow revenues but at a more moderate pace.
Insight #4: A new paradigm for computing is here: Edge Computing
Thanks to the proliferation of sensors and devices, a new paradigm of computing is taking shape - Edge Computing. Edge computing is where data is stored at a device/sensor closer to where it is needed - think your smartphone, smartwatch or your smart temperature sensor. These smart devices in homes or factories are intelligent enough to do data processing and store data. The advantages of edge computing vs the cloud are two-fold: (1) decisions can be made more quickly on these edge/smart devices nearer to end users and (2) it is more economical to process data locally on these edge devices/sensors than sending huge volumes of data to the cloud.
The number of edge devices (including consumer and enterprise) are projected to grow to 29 billion by 2030. All these new devices will take their share of computing workloads that would have otherwise gone to the cloud.
To close: the dynamic nature of technology makes it tough to determine which new forms of computing will take shape in the next ten years. But if we use history as a guide, both existing forms of computing (e.g., mainframes) will become stickier and new forms of computing will arrive – all dampening the path to astonishing growth that public cloud players have seen.
Considerations for investing for slowing IaaS growth
If one believes in the story of slowing Cloud IaaS growth, one may be best advised to stay away from pure plays on Cloud IaaS growth. Thus, another approach, as it is tough to predict how much slowdown Cloud IaaS will experience and how it will impact the market prices, it may be prudent for investors to hedge their risks and consider investment in enterprise software companies (SaaS) that have established businesses in both public cloud and non public cloud environments. As earlier discussed SaaS companies have higher margins vs IaaS. Some examples of these businesses which have healthy on-premise businesses (as each of them started as on-premise) and also have momentum in their cloud businesses include MongoDB ($MDB), Elastic ($ESTC) and Oracle ($ORCL). There is a longer list of enterprise software vendors which fit this profile, these three are included as examples for consideration.
And finally, remember all predictions are often futile.
Peter Lynch: “I deal in facts, not forecasting the future”.
Disclaimer:
This blog is solely for informational purposes and is not intended to be a solicitation to buy or sell any particular security. We suggest you check with a broker or financial advisor before making any investing decisions.
The author is affiliated with Tilden Path Capital, registered as a Registered Investment Advisor in California.
Past performance is no guarantee of future returns.
All investing involves risk and possible loss of principal.
In our analysis, we use third-party resources we believe to be reliable, but cannot guarantee the accuracy of such information. We believe the information we provide is accurate as of the date furnished, but we do not revisit past material to update it.