Wikibon's 2018 Cloud Markets and Trends Report

Premise. Cloud has become the default choice for greenfield applications and systems. However, cloud also is less uniform than it was a few years ago, as customers seek to serve specialized technology needs and vendors seek to serve those needs with competitively defensible positions. Multiple factors will influence the evolution of the cloud market, but data realities, operating scale, business model transitions, edge computing maturity, and geopolitical issues are likely to play a dominant market role.

With Ralph Finos and David Floyer

In 2016, cloud computing options accounted for 8.5% of global enterprise IT spending. In 2017, cloud captured 11.3% of same. By 2027, cloud computing will account for 39.8% of total enterprise IT spending, a CAGR of 17% from $174B in 2017 to $814B in 2027. In other words, Wikibon believes that global cloud computing spending will grow roughly three times faster than global GDP for the next decade.

Cloud is the standard for modern computing. It’s the focus for strategic investment, invention, and business innovation in the tech industry, but also other information- and IP-intensive industries as well. While digital-native start-ups have captured much of the attention, incumbent companies (as IBM calls them) are moving aggressively to adopt digital business strategies centered, at least in part, on exploiting cloud computing.

However, focusing on cloud-based systems is not, by itself, a recipe for success, as companies like GE are discovering. Wikibon believes a more basic trend is in play: The move to treat data as an asset. Digital businesses are transforming work streams, organizational forms, and engagement models around data, often with only modest certainty about strategic outcomes. As a result, firms are adopting technologies – including cloud – practices, and skills that place a premium on strategic empiricism, iteration, and opportunism. While cloud is a baseline for these changes, its expansion is closely aligned with the adoption of a strategic focus on customer experience, Agile methods, AI-infused operations, and a general transition from product-based to service-based business models.

Despite cloud’s impressively strong growth outlook, it does face a number of headwinds. Data assets are easily copied, shared, and corrupted, highlighting serious information security concerns, especially as high-speed networking reaches every corner of the globe. Relatedly, data privacy is increasingly an issue, as communities chafe against corporate and government efforts to aggregate and monetize data with minimal ethical considerations. In part due to differing approaches to data privacy, geopolitical factors are likely to influence the evolution of the cloud market, starting with the EU’s GDPR, but expanding to other domains shaped by cyber competition. Finally, the physical realities of computing, including latency, bandwidth cost, and legacy asset specificities – especially at the edge – are going to play a central role in determining which workloads run where, on what stack, and under who’s control.

Taken together, these forces – and others – will catalyze a cloud technology industry that will:

• Sustain rapid growth. In 2018, Wikibon expects cloud spending will grow 36% to $237B US. Over the next decade, growth will moderate as the cloud industry grows larger, but cloud spending still will grow 17% CAGR, roughly three times faster than global GDP for the next decade, to $814B in 2027.
• Follow the data. The physical and cost realities of data will have a significant impact on the evolution of the cloud industry, especially as edge computing catalyzes billions of additional data sources and locations of system activity. Ultimately, Wikibon believes that the dominant model will be to move cloud to the data, and not move data to central clouds. Consequently, our forecasts call for strong growth in IaaS (15.2% CAGR through 2027) and SaaS (12.8% CAGR) markets, but strongest growth in true private cloud (TPC) options that operate on-premise, near-premises, and at the edge (31.6% CAGR).
• Deeply embed into all businesses and industries. Cloud is a foundational technology for digital business, but the more a company employs cloud services (especially in its engagement model), the more it acts like a tech company. This makes the vendor/user relationship complex. Amazon and Alibaba, for example, are redrawing the line between essential supplier and fierce competitor. Businesses always have had to address fluid relationships, but the fungibility of data makes the situation extreme. The drive to achieve digital scale – now – is forcing businesses to adopt cloud operating models beyond their IT comfort zones. Competitive opportunism, catalyzed by cloud, is reshaping global business. Thus far, governments have applied a light touch in response, but that is likely to change.

The Cloud Market Will Sustain Rapid Growth

In just over 10 years since AWS combined virtualization, high-speed data networks, software-defined resource management, and an online services approach to engagement, cloud has completely reset the enterprise computing norm. Today, almost all enterprise IT organizations – indeed, most businesses – have adopted a “cloud-first” mentality. Every business, everywhere, at every scale is employing cloud in some capacity. The result is an unprecedented wave of technology substitution that is affecting all tech companies and, to some degree, all other industries.

However, unlike other periods of significant technology substitution, the distinction between technology invention and innovation breaks down in cloud computing, with great consequences. Invention is an engineering act: it’s the process of learning something about the physical or social world and turning it into hardware and software. Innovation is a social act: it’s the process of adopting innovation and altering the asset and social relationships that govern work and play. Traditionally, tech – and enterprises in other industries – invented something and packaged it as a product with intrinsic value. Marketing, selling, and service functions, in concert with channel and customer adoption, were the basis for innovation. These processes determined if a product generated value in use; ultimately, if it were adopted.

The relationship between invention/innovation in the cloud experience, however, is mushed. Cloud is a service. By design, users pay only for the services they adopt and use. This fundamentally alters the seller/buyer relationship, and therefore the functional relationships within a business.

The seller/buyer relationship is changed because exchange now must include significant volumes of data: data that provides the cloud services and data that monitors and determines if cloud services have been appropriately delivered. Because data is fungible, this opens the door to new types of competitive opportunism. While technologies like blockchain are intended to address this opportunism, the tech industry is still working through crucial business model, IP protection, trust, and engagement security issues.

Functional relationships within a business are changed because many customer adoption aspects of marketing, selling, and service functions are obviated by cloud, once the cloud service relationship is established. Service capabilities can be added or enhanced with modest, and sometime no, customer effort. The impacts are significant. We’re in the midst of an enterprise computing landgrab as cloud technology providers seek to firm up enterprise customer relationships for long-term advantages. However, like any profound technology transition, this one is messy. Some cloud companies, for example Salesforce, are demanding longer-term contracts on day one, making their “no software” services look a lot like enterprise software contracts. However, while cloud business and relationship models will warp as buyers and seller gain experience, cloud has two important catalysts for continued success: (1) scale efficiencies provide cloud players a lot of strategic and pricing options; and (2) technology invention increasingly is centered on cloud models.

As a result, Wikibon believes that the cloud market will:

• Grow 17% through 2027. In 2017, enterprises spent $174B on cloud technologies and services (see Figure 1). By 2027, Wikibon believes that figure will grow to $814B (in current dollars), a real CAGR of 17%. Our market models show that the largest cloud vendors in this time frame will be pushing $100B in annual cloud revenue alone. However, while the largest hyperscalars in aggregate will constitute the largest block of IT buying (as the basis for their cloud services), and market concentration will increase, cloud spending by 2027 still will account for less than 40% of all enterprise technology buying in 2027. The technology industry is not headed for monopsony.
• Remain dominated by SaaS. The largest segment of the cloud industry is SaaS, and it will remain so. In 2017, the global SaaS market was $104B. By 2027, we believe it will grow to $346 billion, a CAGR of 25.9%. However, this is the most conservative portion of our forecast. The transformation to digital business, combined with the fungibility of cloud, will spawn an unprecedented period of business specialization. The range of SaaS businesses is likely to explode, as new business forms, operating models, and regulatory regimes shape new value propositions and engagement options.
• Expand specialized service options. The cloud model is the driving force for most of the tech industry’s invention today. Software-defined everything, hyperconverged infrastructure, ARM-based architectures, distributed data management, cybersecurity, advanced analytics, and AI are all examples of technologies catalyzed by cloud concepts or opportunities. That will continue, even accelerate. As recently as two years ago, Wikibon predicted that cloud companies (notably AWS) would have to reset approaches to rolling out new services. But what we (and others) thought was a bug in the cloud business model is turning out to be a feature. AWS, Google, and Microsoft, in particular, are ramping the rate of new service introductions and enterprises – and, notably, ecosystem partners – are showing no signs of fatigue. Usability and simplicity will remain important, but as digital business requirements expand, demand for service specialization will increase.
• Support multiple business model forms. The aforementioned landgrab really is a push by cloud companies to get as much “data under management” as possible. Consequently, the dominant business model has been (1) get data into our public cloud; then (2) provide services against that public cloud-resident data. However, this “data first” model isn’t the only possible model. “Service-first” models (e.g., Oracle’s Cloud-at-Customer) are maturing and are likely to gain traction, at least in specialized segments. Moreover, cloud ecosystems are using digital business capabilities to rewire entire asset, product, and customer lifecycles, both in the traditional tech business and non-traditional tech businesses. As data reduces asset specificities in all industries, new strategic options are created. Digital business transformation is impacting the tech industry via the cloud, but also all other industries, as well.

Figure 1. Total IT Spending ($B) 2017-2027. Source: Wikibon

Figure 2. Public Cloud Spending $B 2017-2027. Source: Wikibon

The Cloud Will Move to the Data

In his original post about data gravity, Dave McCrory posited that, “As Data accumulates (builds mass) there is a greater likelihood that additional Services and Applications will be attracted to this data.” While McCrory noted that services and applications would move to the data, many interpreted the concept to first mean that more data would move to the data, which would result in all data being located in “centralized” public clouds. As a consequence of this thinking, many predicted the end of local processing, data centers, and – generally – IT organizations.

Wikibon has never subscribed to the “first, data moves to the cloud” proposition. Rather, we subscribe to McCrory’s original notion: services and apps, in the form of the “cloud experience,” are moving to the data. We proposed that a True Private Cloud (TPC) market segment would emerge, comprising on-premise, near-premise (e.g., local co-location), and edge computing resources, that provided a common, service-oriented, fungible – and data first – approach to simpler, cloud-based IT.

Our research confirms the emergence and growth of this segment. Based on our user conversations, IT organizations are adding TPC to their cloud strategies in direct response to concerns about data latency, costs to move, regulatory concerns, and IP protection. Moreover, public cloud companies (e.g., AWS and Google) are adding TPC-like services (with more coming); indeed, Microsoft is using the on-premises Azure Stack to differentiate its cloud strategy (with modest uptake to date, but our research suggests momentum is gaining). Systems companies finally are offering (sometimes rudimentary) pay-as-you-use options to their converged infrastructure (CI), hyperconverged infrastructure (HCI), and other infrastructure offers (e.g., Oracle, HPE, IBM, Dell EMC). And near-premise hosting and co-location firms (e.g., CenturyLink, Expedient) are adding locations and advanced cloud capabilities to their service portfolios.

Taken together, moving the cloud to the data will push the TPC segment ($186B) to be larger than the IaaS segment ($176B) by 2024 (see figure 3). Because TPC will grow considerably faster than IaaS (29.2% CAGR TPC; 15.2% CAGR IaaS), the revenue gap between the two will widen. By 2027, the TPC market should top $262 billion globally, whereas the IaaS segment will grow to about $206B.

Four domains will be especially important to shaping the arrangement of cloud resources, each for very different data gravity reasons. Wikibon believes the cloud will:

• Reach the edge. The most misleading term in the computing industry today is “fog computing,” which suggests that edge computation will be ephemeral and amorphous relative to the clarity of the cloud. On the contrary, edge computing will require unprecedented technological precision. Real-time, near-perfect representational fidelity (e.g., “digital twins”), environmental instability, inherent vulnerability – these are just a few of the physical challenges presented by the edge. Add the “softer” challenges of life and death, ethical agency, consensus reliability, and privacy and it becomes clear that edge computing will be a driving factor of change in many domains, including cloud. Our models suggest that data generated at the edge will tend to stay at the edge, leading to new approaches for shipping function to edge resources. For edge environments that require a degree of plasticity (often called “mid-edge”), true private cloud systems will help. For edge systems that require programmability at smaller scale, serverless computing will dominate. The cloud’s virtual foundation will have to evolve to extend to the edge, but the edge and the cloud will complement.
• Absorb the legacy. For thirty years, commentators and vendors have been suggesting that legacy applications would be retired, and systems migrated to more current platforms using modern tools. And yet, these high value traditional applications (HVTA) persist. Predictions that HVTA will begin moving to public cloud platforms are widespread, but the risks of migration remain untenably high for the vast majority of businesses. Does that mean that nothing will happen? No. Wikibon believes legacy applications represent an important true private cloud opportunity. We estimate that the software legacy could easily be costing global businesses $1.2 trillion per year. While re-platforming these applications won’t completely eradicate these costs, bringing them beneath a cloud umbrella using true private cloud technologies integrated into legacy platforms will allow businesses to better manage operating costs and enhance leverage opportunities while incurring minimal migration risks.
• Dominate AI innovation. Many of the algorithms that are the basis of AI have been available for years. However, four key technological factors now make AI possible: (1) flash-based systems that are engineered to rapidly and simply deliver large volumes of data; (2) low-cost GPUs that are especially suited to processing AI data models; (3) ubiquitous devices and sensors generating nearly 3 exabytes of real-world data daily; and (4) the cloud, which has the scale to aggregate that data for AI model building, training, and maintaining. Big data began a move to the cloud in 2017, and there it will stay. AI and other data-first application technologies are following and will be a primary driver of public cloud (i.e., SaaS and IaaS) demand. While model training will occur in the cloud, model inferencing will occur largely at the edge.
• Settle on multicloud architectures. Most enterprises use multiple clouds today. The question is how fast can enterprises converge those clouds into cohesive digital business platforms capable of efficiently supporting current needs and creating future business options. Wikibon believes that this will require enormous new invention, but that the industry is capable of creating the new technologies and services required. In particular, we point to the emergence of open source as the dominant framework for creating new software. As cloud vendors provide new services, those service become clearer targets for the open source community. This won’t lead to a renewed focus on base cloud infrastructure, like OpenStack, but instead be directed at multicloud technologies and tooling like Kubernetes, Istio, and others. A rethinking of many IT practices will be required, including adopting a microservices approach to capability isolation, but we don’t foresee any cloud vendor being able to close the cloud – especially if Google and IBM continue to push open source technologies as the basis for cloud invention.

Figure 3. True Private Cloud Spending ($B) 2017-2027

Cloud Ensures Digital Business Transformation Impacts All Industries

Arguably, today’s tech industry changes are the most profound in history. Previous generations of technology change were significant, to be sure, but generally they followed a “known process, unknown technology” pattern. For example, we knew we were going to substitute technology for labor in accounting activities, we just didn’t know which technology we would use, how reliable it was, how much it would cost, etc. Thus, the focus tended to be on the classes of infrastructure that featured in the change, tracking technology platform evolution from mainframe to web. Enterprises tended to focus on the savings that could be generated by substituting different classes of technology for different forms of operational activities, like accounting, HR, etc. CIOs and IT leaders could deliver 5-8% cost savings for a given mix of application delivery by leveraging Moore’s Law-based price/performance improvements.

Today’s relationship between technology and business is very different. Known processes are already “eaten by software.” While enterprises are constantly refining ERP and other operational applications, attention is turning to use cases that don’t feature well-defined process models. However, while processes are less well known, cloud will be the technology base. Thus, today’s strategic applications can be characterized as “unknown process, relatively known technology.” Moreover, as limits to Moore’s Law (or, perhaps more accurately, Denard scaling) solidify, CIOs need to reset expectations for hardware productivity improvements – and IT budget impacts.

Technology is more embedded within business than ever before, and that catalyze greater enterprise attention on:

• Digital business transformation. Amazon, Netflix, Google, Apple and other digital heavyweights have shown the way: turning data into assets that can advantageously alter a company’s value proposition, organizational relationships, and patterns of work is the basis for better customer experience, profitability, and improved valuation. Indeed, digital business transformation is best thought of as the process by which a business adopts and exploits data-as-an-asset opportunities. Every business is being impacted – regardless of size, industry, and location(s). The relationship between digital business transformation and cloud adoption is data: A company’s arrangement of data assets generally will dictate its arrangement of cloud resources.
• Transitioning product-based to service-based business models. One of the transformative elements of digital business is the use of data to alter the fundamental rules of business interactions. Traditional notions of business are oriented to “value-in-exchange”: the proposition that a customer purchases something valuable and then is responsible for liberating it in their business. One-time payments occur up-front; after the sale, engagement is limited to support. Digital businesses, though, emphasize notions of “value is use,” which emphasizes the customer’s outcome and how a good or service does or does not help achieve it. Subscription payments occur as an offering is used; engagement is continuous. Data is essential to value-in-use business models. It’s the basis for economically sustaining engagement, measuring utilization, and minimizing disputes. The cloud is the technology industry’s service-dominant business model – and essential to generally supporting service-dominant business models. It’s the basis for ubiquitous digital endpoints, flexible capacity, and handling hypervolumes of data. Cloud demand will accelerate as more industries shift to service-dominant business models.
• Exploit multicloud operating models. The CI/CD, Agile development, microservices-based solutions, and pay-as-you go operating models of the cloud are diffusing and being adopted in enterprises generally. While these practices aren’t anywhere near as keen in typical enterprises as they are in cloud leaders, experience and tooling eventually will close the gap. Our research suggests that CIOs and business leaders are focused on establishing unique multicloud operating models that serve their particular business’s needs, using customer experience, HVTA legacies, institutional, and IP concerns as guideposts. Indeed, our forecast assumes that these considerations are crucial to long-term enterprise cloud strategies and that multicloud operating model tooling and experience will mature rapidly over the next 5-7 years.

Three Scenarios: Dull Edge, Data-In-Place, and Cloud Inversion

Forecasting is tricky. Forecasting during periods of turbulence is especially tricky. In this report, we’ve detailed the trends likely to shape growth in the cloud industry, but much remains uncertain. Our assumptions regarding business changes, technology invention, and geopolitical factors are reasonable, but still subject to wild swings in fortune. Wikibon believes the public cloud business will grow 13.2% CAGR through 2027 and that TPC over the same period will grow even faster (albeit off a much smaller base) at 31.6% CAGR. However, contingencies certainly will emerge. For example, we assume that HVTAs will migrate to public clouds with great difficulty. However, if AWS and other data-first cloud vendors can introduce really advanced tooling for streamlining database and other traditional technology migrations, that would have a great impact on enterprise cloud strategies.

Without offering detailed market figures, we think three alternative trend and growth scenarios are worth mentioning. They are:

• The Dull Edge Scenario: Public Cloud Grows Faster. Wikibon believes that the edge will have a major – even dominant – effect on future industry growth. Our expectation is for more intelligent devices generating more data with increasingly powerful inferencing capabilities shaping increasingly real-time automation. However, if the edge doesn’t “sharpen,” it won’t generate significant need for TPC options. This could happen if, for example, operational technology (OT, currently the function responsible for SCADA-oriented edge systems) and IT can’t unify regarding edge needs, AI ethical concerns limit deployment of complex automation, volume-oriented devices and management software can’t accommodate edge security requirements, or edge management systems capable of handling real-time requirements don’t mature. Under this scenario, public cloud growth accelerates and TPC growth slows. The surest marker that this scenario will emerge is continued tension between OT and IT.
• Data-in-Place Scenario: TPC Grows Faster. Presuming the edge sharpens, this scenario presumes an acceleration of the cloud model moving to the data. At its core is the observation in many of today’s cloud vendor business models, the presumption is that customers will move data into public cloud resources, and then employ cloud-based services on that data. This “data under management” approach is favored by AWS, Google, and Alibaba and the big SaaS players, including Salesforce. However, our research suggests that large enterprises, in part bolstered by conservative HVTA strategies, will pursue hybrid cloud strategies that increasingly favor cloud vendors that offer high-value cloud services to most data locations. Serverless computing, for example, provides this capability. For this scenario to emerge, multicloud management, multicloud integration, and multicloud data protection all will require significant new invention, essentially closing the cloud operating model gap between public and TPC options. Microsoft, Oracle, and IBM all are pursuing this strategy, bolstered by the commitments of Dell EMC, HPE, and Cisco. The surest markers that this scenario will emerge are the continued rise of serverless computing and advances in multicloud management systems from the TPC players and their allies.
• Cloud Inversion Scenario: The Cloud Substitution Slows. Cloud’s benefits are uncontestable, but digital business’s benefits can be questioned on multiple ethical, business model, and geopolitical grounds. For example, communities may reject greater automation, luddite movements may slow AI adoption, or politicians may conflate sovereignty with cloud ownership. While Wikibon is not predicting any of these factors, conversations regarding each of them – and many others – are getting more vociferous; they can’t be completely discounted. Should combinations of these factors come into being, cloud substitution will slow as service inventions are more scrutinized and innovations slower to be adopted. The clearest marker for this scenario is GDPR, which may have started as a sincere effort to enact stricter privacy laws for EU citizens, but could morph into a political tool for undermining the expansion strategies of US-based and China-based cloud companies.

Action Item. At this point, every enterprise should be pursuing a cloud-first IT strategy. However, cloud-first should include bringing cloud capabilities to edge and HVTA systems. Increasingly, multicloud architectures should be predicated on bringing cloud services and innovation to data. CIOs should be establishing pragmatic approaches to building the multicloud operating model capabilities required by their enterprise to succeed in an increasingly digital world.