According to IDC, this year global spending on public cloud services will exceed $1 trillion for the first time – an increase of more than 21%. The market will double by 2029. Cloud infrastructure spending overtakes traditional on-premises investments, and enterprise landscapes are almost always a combination of public, private and hybrid cloud.
Change of image
The main change in recent years is a change in perception. Previously, the cloud belonged to the category of technologies: new, not fully understood, requiring pilots and careful implementation. Today this stage is completed.
Clouds have existed for decades, but it is the last 5-7 years that have become a period of mass adoption. During this time, standards of use were formed, architectural approaches were established, market leaders appeared. Almost any business today is digital, that means it directly depends on IT.
If IT is the infrastructure basis of business, then the cloud becomes the infrastructure basis of IT. Today it is the default environment for corporate architecture.
What became the driver
The transition of the cloud to the status of a full-fledged infrastructure is the result of the simultaneous action of three factors.
The first is technological maturity. High penetration rate, emergence of leaders (vendors and platforms), formation of de facto usage standards. The market has accumulated enough cases to remove the most pressing issues and reduce uncertainty.
The second is economies of scale. With the growth of the market and competition, cloud solutions have become cheaper. Economies of scale, optimization of provider infrastructure, and the development of a service model have made the cloud economically viable in a wide range of scenarios.
The third is business pressure. Companies are no longer prepared to wait months and years. They need speed, flexibility, and the ability to scale quickly. The first successful cases gave the market proof, and this led to an avalanche-like increase in demand.
Cloud evolution
Ten years ago and today clouds are two different entities.
The early model was simple. Virtual machines, basic IT services, partial use, low integration into the corporate landscape. Companies brought small volumes of tasks to the cloud, the requirements for providers were just as simple.
But business was digitized and IT inside it became more difficult. The number of systems, microservices, and critical processes grew. There are new requirements for fault tolerance, for the complexity of applications. Simply renting a couple of virtual machines has become insufficient.
Today’s infrastructure is heterogeneous. Not all companies take everything to the cloud – critical information systems remain on-premises. At the same time, many other systems and microservices live in the cloud, and close interaction should be ensured between them. This is how hybrid installations appeared.
S3 storages are used for developers and backup tasks, and managed services like Kubernetes are used for microservice architecture. All this must be combined into a single ecosystem not only on the part of the provider, but also on the part of the customer, so that it is convenient for him to synchronize the entire set of systems and applications.
New normal
Complete abandonment of local infrastructure is rare. Some systems remain critical and require control at the company level. So Legacy systems have not gone away.
As a result, the hybrid scenario has become the norm. Local infrastructure remains in the customer’s perimeter, the cloud takes on scalable, dynamic loads, and communication is built between them.
In parallel, the platform level is developing: Kubernetes, PaaS, managed services allow to work not with the infrastructure directly, but with a higher level of abstraction. The cloud covers not only infrastructure tasks, but also the tasks of development, operation and analytics.
Speed is the new main factor
One of the most painful changes in recent years is the decreasing planning horizon. Previously, projects were planned for years, today – for quarters and half a year. Businesses need to test hypotheses quickly, launch products, and adapt to the market.
Under these conditions, the classic equipment procurement model has become a bottleneck. The standard delivery time is 12-15 weeks. Due to the complexity of parallel imports and secondary sanctions, it can be safely multiplied by two, and equipment can be confiscated or arrested when crossing the border.
As a result, the project for the construction of on-premises infrastructure is delayed for six months or more. Such deadlines are unacceptable in any organization, and it becomes impossible to plan a financial model.
The cloud removes this problem. Infrastructure deployment takes days to weeks depending on scale. The company does not bear any quantitative and qualitative risks.
Cloud perception errors
Despite the maturity of the market, many companies continue to take the cloud in a simplistic way.
The first and main mistake is to assume that the cloud “is the same servers, but not inside us.” In reality, this is a different architecture and a different logic of consumption.
The second mistake is the myth that the cloud is slow and complex. In fact, the cloud, on the contrary, is an acceleration tool. Example: you have a project that is given a month, and you know that there will be an influx of users. In the old paradigm, we would have to urgently look for equipment and urgently raise IT resources based on it. In the cloud, an infrastructure reliably calibrated for a task will be deployed in a few days.
The third mistake is an attempt to transfer everything exactly as it was. The cloud requires a rethink of architecture, not just moving data around.
Fault tolerance instead of own hardware
Ownership of “hardware” gives way to ownership of fault tolerance, availability, data safety. You can own a physical server, but not have fault tolerance.
If this is one unprotected server that will fail at any time and lead to a loss of revenue, there is no use from direct ownership. The real value is in business continuity and data integrity, and the cloud provides just that model.
Vendor lock-in
Until recently, linking to a provider was one of the main barriers, but today the situation has changed. The modern strategy of any mature cloud provider is not to lock the client technologically, but to keep the quality of the ecosystem.
There is a large amount of software on the market for converting from one hypervisor to another. Moving between clouds has become quite simple and fast. For databases and managed services, no one forbids moving either: the client can re-deploy the service on standard images if he has the necessary competencies.
Real economy
The cloud is not always universally beneficial.
It is especially effective in conditions of uncertainty, short planning horizon, absence of CAPEX and when fast start is required. On large infrastructures and with a long ownership horizon (3-5+ years), own on-premises infrastructure can be cheaper. If, in addition, the company transfers absolutely all IT tasks to the provider, the cloud becomes more expensive.
However, recent events have changed the alignment. A server that cost $15,000 costs $30,000 today. The equipment went up instantly and avalanche-like. Companies are not ready to bear such capital costs. What was budgeted in November is no longer relevant in February.
Therefore, the real strategy of many companies today looks mixed. At the start, everything is in the cloud, and then there is optimization, redistribution of loads, return of some of the on-premises components, and in the final – a hybrid model, where the cloud remains more for dynamic and urgent tasks.
Constraints and stop factors
The main limitations lie in the regulation. Public sector and the financial sector have strict requirements for data localization. Information security departments of such organizations often do not allow sensitive data to be transferred to the cloud according to internal regulations.
Although everything is technologically ready, the problem is now rather in the legal and bureaucratic aspect. Gradually, banks are beginning to be allowed to go into the clouds, which means that the public sector will also move in this direction.
Information security risks are often perceived as a barrier, but in practice, mature cloud providers provide a level of protection comparable to or greater than local solutions. Information security processes for a reliable provider are paramount, because reputational and economic losses from leakage can lead to bankruptcy.
Cloud opportunity reassessment
The greater the amount of delegation, the more project costs inflate relative to total cost of ownership. This is also true for large infrastructures over a long horizon.
However, companies often need speed and competence. Large business knows how to count money and goes to the cloud consciously, realizing that on the horizon 1-2 years it will overpay, but will close the internal deadlines and strategic tasks. Then it gradually refuses some of the services (for example, from managed IT), acquires its own competencies, rebuilds processes and comes to a hybrid scenario, where some piece remains in the cloud.
Ultimately, cost effectiveness is achieved.
Russian specifics
The local cloud market is several times smaller than the global one. However, at the same time, the Russian economy is very digitized. Public services are developed; the level of online services is high. The business is spoiled by ready-made services and wants to solve problems not in a week, but in an hour.
According to iKS-Consulting, in 2026 the volume of cloud services in Russia will reach 530 billion rubles, and by 2030 – 1.2 trillion rubles.
The structure of consumption will differ from foreign one. Firstly, a large number of government agencies in the future 5-10 years will move to the cloud; that introduces special requirements for the safety and integrity of data. Secondly, due to the high degree of digitization, the demand for highly specialized services will grow – for example, graphic calculations for specific tasks, and then ready-made AI models for specific business cases (LLM training for car sales, for technical support of a telecom operator, etc.).
The service delivery form will move to the next level of abstraction.
In addition, the number of local software and hardware vendors is growing, the technological stack is changing. Architecturally, the cloud will not change much, but internal technologies will be modified relative to foreign counterparts.
The future of cloud and infrastructure
The current trend is the transition from IaaS to platforms, and from platforms to AI services. GPU as a service today is already a standard request. In the coming years, we will see the transformation of IaaS into an analogue of platform services for AI tasks.
The next stage is the infrastructure “by button” or “by request.” The boundaries between infrastructure, platform and application will be erased. A ready-made piece of services necessary for the functioning of the business will be deployed under a text or voice request. This is futuristic, but quite realistic on the horizon of 5-10 years.
Cloud is no longer an alternative to infrastructure. This is the new infrastructure. For CIOs, this means a change of focus: the question is no longer whether to use the cloud, but how to build the architecture effectively in a cloud paradigm, managing the speed, cost and sustainability of the business.
By Alexey Makarkin, DataSpace Cloud Product Director