What is the cloud market worth in 2026?

Cloud computing is no longer an emerging trend – it is the default substrate of business IT, and the money proves it. Gartner forecast worldwide end-user spending on public cloud services at roughly $675 billion for 2024, rising toward the mid-$700-billion range in 2025 and continuing to climb into 2026. Cloud is now one of the largest categories of technology spending on earth.

But the most interesting statistics of this decade are not about growth alone. They are about correction: what cloud actually costs, what gets wasted, and why a meaningful share of workloads is moving back off the public cloud.

How much public cloud spend is wasted?

• Organizations consistently estimate that more than a quarter of their cloud spend is wasted – idle resources, oversized instances, and forgotten services. Flexera’s State of the Cloud research has put self-reported waste near 27–30% year after year.
• In the same research, managing cloud spend has overtaken security as the number-one cloud challenge reported by organizations.
• For a business spending $20,000 a month on public cloud, a 27% waste rate is roughly $65,000 a year paying for nothing.

Variable, consumption-based billing is the root cause: costs move month to month, egress fees punish data movement, and nobody’s job is to turn things off. This is the single strongest argument for predictable, fixed-cost infrastructure – and it is why cost predictability has become a board-level topic.

What is cloud repatriation – and how common is it?

Cloud repatriation is the movement of workloads off public cloud and back onto private cloud or company-controlled infrastructure. It has moved from heresy to mainstream:

• IDC research in 2024 found that roughly 8 in 10 IT decision-makers expected to repatriate at least some compute or storage within the following year.
• A widely-cited 2024 Barclays CIO survey reported 83% of enterprise CIOs planning to move at least some workloads back to private environments.
• Uptime Institute surveys have found about a third of organizations have already moved workloads back from public cloud.
• The most famous documented case: software company 37signals left AWS and reported saving about $2 million in the first year, projecting roughly $10 million over five years after buying its own hardware.

Repatriation is not anti-cloud. It is the market maturing: steady, predictable workloads are often dramatically cheaper on private infrastructure, while burst and experimental workloads still favor public cloud. The winning architecture is deliberately hybrid.

Why are private and hybrid cloud growing?

• Cost control: steady-state workloads on owned or private-cloud infrastructure carry fixed, predictable costs – no consumption meter, no egress surprises.
• Data sovereignty: regulators and clients increasingly ask where data lives and who controls it. Roughly twenty U.S. states now have comprehensive privacy laws, and sovereignty requirements in regulated industries keep tightening.
• AI changed the calculus: the AI buildout made GPU capacity, data governance, and confidentiality central to infrastructure decisions. Businesses that want AI on their own data increasingly want that data – and the models touching it – on infrastructure they control.
• Hybrid is already the norm: Flexera’s research consistently finds the overwhelming majority of enterprises operating multi-cloud and hybrid estates rather than a single public cloud.

What do these statistics mean for small and mid-sized businesses?

The pattern the Fortune 500 discovered at $100 million scale applies at 50-employee scale: match the workload to the platform, and make the costs predictable. In practice that means a hybrid design – private cloud for the steady core (line-of-business applications, data, voice), public cloud where elasticity genuinely pays – delivered at a fixed monthly fee so the meter never runs.

That has been Infradapt’s model since long before it was fashionable: we have operated our own private cloud infrastructure for client workloads for more than fifteen years, offering private and hybrid cloud services where your data, applications, and voice run on systems we own and operate – with true single-tenant privacy and fixed-fee economics. The same philosophy now extends to managed private AI: your data and AI workloads on controlled infrastructure, governed and predictable.

Sources and further reading

Figures above are drawn from published research and reporting by Gartner (public cloud spending forecasts), Flexera (State of the Cloud reports), IDC (workload repatriation research), Uptime Institute (capacity and repatriation surveys), Barclays (CIO survey coverage), and 37signals’ public accounts of its cloud exit. Statistics are the sources’ own; refreshed on this page as new editions publish.

Talk through your own cloud numbers

A 30-minute conversation maps your workloads to the right platforms – and shows what fixed-cost infrastructure would mean for your budget.

Talk to a Solutions Specialist

Or call 1-800-394-2301. Serving businesses across Eastern Pennsylvania and New Jersey from offices in Allentown and Philadelphia.