Cloud customers should prepare for a new wave of infrastructure inflation that may affect cloud bills as early as 2026, according to Octave Klaba, CEO of European cloud operator OVHcloud. Klaba stated in a recent post on X that the business anticipates a 5-10% increase in the cost of several cloud products between April and September 2026, mostly due to rising memory and storage costs throughout the global hardware supply chain.
At the heart of the problem is an aggressive reorientation of semiconductor production toward AI workloads. Manufacturers are prioritizing high-margin memory used in GPUs and accelerator platforms, reducing available capacity for “standard” RAM and NVMe components used in general-purpose servers. That capacity squeeze is now feeding directly into pricing.
Klaba highlighted that prices for RAM and NVMe drives are set to increase significantly within roughly six months, with consequences that will extend well beyond hyperscale clouds and large SaaS vendors. Any business that depends on servers – whether for public cloud, private cloud, colocation, or on-premises infrastructure – is likely to feel the impact as new hardware is procured or existing fleets are refreshed.
Independent market data aligns with this outlook. Analyst firm TrendForce reports that since September 2025, DDR4 prices have climbed by 158%, while DDR5 modules have surged by 307%. At the same time, large vendors such as Samsung have already raised some memory prices by around 60%, underscoring how quickly conditions are shifting across the memory market.
According to CEO Octave Klaba, OVHcloud’s internal projections indicate that a server produced in December 2026 could cost between 15% and 35% more than an equivalent configuration assembled in December 2025. For cloud providers operating at scale, those increases translate directly into higher capital expenditure and, ultimately, higher service prices, particularly on compute and storage offerings that are most sensitive to component costs.
Revisiting Hybrid Strategies
In the short term, parts of the industry are attempting to cushion the blow. Mr. Klaba notes that the global supply chain is pulling demand forward by ordering components roughly six months earlier, at today’s lower prices. That strategy may allow cloud providers and OEMs to manufacture servers for an additional 6–12 months at costs below what the “true” market price will be once the new AI-driven pricing fully flows through.
However, this defensive move has a double edge. While it extends the period of relatively lower pricing, it also accelerates demand for components today, which in itself would contribute to upward price pressure starting from December 2025. The result is a temporary buffer for production pipelines, not a structural fix.
Between April and September 2026, OVHcloud expects to pass through a portion of these cost increases to customers, in the form of 5–10% higher prices on some cloud products. During that same timeframe, server costs are estimated to be up by 15–25%, depending on configuration and component mix. Mr. Klaba emphasizes that these figures are estimates based on information available as of November 2025, and that the situation could move faster if AI demand intensifies or if additional supply disruptions occur.
Despite the looming cost pressure, analysts do not forecast a broad retreat from cloud computing. While some enterprises may revisit hybrid strategies or selectively repatriate workloads to on-premises environments, the overall direction of travel remains toward cloud-centric architectures. For many CIOs and CTOs, the conversation is likely to shift toward cost optimization, workload placement, and contract negotiations rather than a fundamental rethink of cloud as a delivery model.
For B2B buyers and technology leaders, the message is clear: memory and storage are becoming strategic variables in cloud economics, heavily influenced by the trajectory of AI adoption. Organizations planning large migrations, refresh cycles, or AI infrastructure rollouts over the next 12–24 months may need to revisit their TCO models, budget assumptions, and timing, as hardware-driven price dynamics increasingly reshape the cloud cost landscape.
Executive Insights FAQ
What exactly is driving the anticipated 5-10% increase in cloud prices?
The primary driver is a sharp increase in hardware costs, especially RAM and NVMe storage. Semiconductor manufacturers are prioritizing high-margin AI-oriented memory for GPUs, which constrains supply for general-purpose components. As the cost of building servers rises, cloud providers are preparing to pass part of that increase through to customers.
How severe are the underlying memory price hikes in the current market?
Recent analyst data indicates that DDR4 prices have risen by 158% and DDR5 prices by 307% since September 2025, while some Samsung memory products have seen price increases of around 60%. These shifts materially affect the bill of materials for every new server deployed into cloud and data center environments.
When should enterprises expect these higher costs to show up in their cloud bills?
OVHcloud’s current projections suggest that certain cloud product prices will rise by about 5–10% between April and September 2026. Component prices, however, start increasing from December 2025, so the exact timing and magnitude for each provider will depend on existing inventory, procurement strategies, and contract structures.
Can enterprises mitigate the impact by hoarding hardware or bringing workloads back on-premises?
Short-term stockpiling of components may delay the impact for organizations that buy and manage their own hardware, potentially extending lower effective pricing by six to twelve months. However, this does not change the underlying market trend and can even add to near-term demand pressure. While hybrid and repatriation strategies remain options, they introduce their own costs, complexities, and timelines.
What should CIOs and CTOs do now to prepare for these shifts?
Technology leaders should refresh their cost models for 2026 and beyond, incorporate higher server and memory costs into TCO analyses, and revisit cloud contracts and reserved capacity commitments. It is also prudent to prioritize workload right-sizing, storage tiering, and architecture optimization, and to align procurement and migration timelines with the expected windows of price change.
