Why enterprise hardware costs and lead times are not coming down any time soon

Something fundamental shifted in the enterprise hardware market in late 2025, and by 2026 the consequences are being felt across every sector and every size of organisation.

If you have recently approached a reseller for a server, a storage array, a networking switch, or even a batch of memory modules, you will already know: the quotes are taking longer, the prices are higher than budgeted, and the lead times stretch further with each passing quarter.

This is not a temporary blip driven by a single disruption. It is the result of a structural realignment of the global semiconductor supply chain, one that began quietly, accelerated rapidly, and is now reshaping the economics of enterprise IT procurement for years to come.

Understanding what has happened, why it happened, and where it is heading is no longer optional. It is the difference between a refreshed, resilient infrastructure and one that stalls on budget approval and vendor allocation queues while the business waits.

The origin: when AI ate the semiconductor industry

The story begins not with a natural disaster but with demand. Specifically, the explosive and largely unprecedented demand for AI infrastructure that swept through the hyperscale cloud sector from 2023 onwards.

The development and deployment of large language models, generative AI platforms, and GPU-accelerated workloads required a category of memory that did not previously exist at commercial scale: High Bandwidth Memory (HBM). Chips like NVIDIA’s H100 and H200 series depend on HBM3 and HBM4 stacked memory to deliver the throughput that AI training and inference demands. Producing HBM is extraordinarily capital-intensive. It also consumes the same fabrication lines, the same advanced packaging capacity, and many of the same raw materials as conventional enterprise memory such as DDR4 and DDR5.

The three companies that control the vast majority of global DRAM production, Samsung, SK Hynix, and Micron, made a rational commercial decision: they shifted their manufacturing capacity towards the high-margin AI memory segment and away from the commodity enterprise memory that has served data centres for years. The consequences for the rest of the market were swift and severe.

Deloitte’s 2026 Semiconductor Industry Outlook captures the scale of the shift clearly. Global semiconductor revenues are forecast to reach approximately US$975 billion in 2026, up 25% year on year, with AI chips expected to account for roughly half of all industry revenue. Meanwhile, the success of that AI boom has created critical shortages in conventional DDR4 and DDR5 memory. Deloitte noted that prices for these products rose approximately fourfold between September and November 2025 alone.

From memory crisis to market-wide disruption

By early 2026, the memory shortage had moved well beyond a component-level inconvenience and was driving measurable pricing increases across the full stack of enterprise hardware.

In February 2026, Gartner issued a stark assessment: surging memory costs were projected to drive a 130% increase in memory prices by year-end, resulting in a 17% rise in PC prices and contributing to a 10.4% decline in global PC shipments. This is the first such decline driven primarily by pricing rather than demand collapse. Gartner also predicted that business PC lifetimes would lengthen by approximately 15% as organisations deferred purchases in the face of inflated pricing.

Micron confirmed publicly that it could fulfil only around 55 to 60% of core customer demand. Western Digital indicated it was effectively sold out of hard drives for 2026. These are not fringe suppliers. They are foundational to the infrastructure stacks that power the servers, storage arrays, and workstations deployed by organisations of all sizes.

The disruption compounded when tariff policy entered the equation. US trade policy in 2025 and 2026 introduced additional cost pressures across hardware manufactured in or routed through certain Asian markets. The enterprise networking segment felt this acutely.

• Cisco updated its compute pricing rules in early 2026, removing established promotional structures and introducing list price increases, with Catalyst Switch pricing reported to have risen by 5 to 10%.
• HPE increased server prices by approximately 8% in response to combined component and tariff pressures.
• Dell announced price increases across commercial PCs, workstations, and monitors.
• Lenovo cancelled all outstanding hardware quotes in January 2026 and reissued new pricing with estimated increases of 10 to 15% on servers and PCs.

The Arctiq analysis of the 2026 IT supply chain crisis captured the procurement reality with precision: quote validity windows from major vendors have shortened dramatically, pricing guaranteed last month may not survive long enough for a purchase order to clear legal review, and lead times for enterprise servers are stretching further every quarter.

Who is affected?

No segment of the market is insulated.

Small and mid-sized businesses, which typically lack the contractual leverage to secure priority allocation from tier-one vendors, are finding that standard orders are being pushed to the back of fulfilment queues. Large enterprise organisations face the challenge of refreshing ageing infrastructure at a time when both budget and availability are constrained.

A survey conducted by Omdia in early 2026 found that more than 90% of channel partners had already experienced changes to vendor fulfilment timelines, and more than 70% had seen price increases passed through to end customers.

The storage market is not exempt. IDC has observed volatility in SSD and HDD pricing driven by the same upstream memory dynamics, and Pure Storage, Dell EMC, NetApp, and other tier-one storage vendors are all navigating extended component lead times that compress their ability to build and deliver systems at the pace customers historically expected.

The outlook: what 2027 holds, and why waiting may cost significantly more

It is tempting to conclude that waiting out this period of disruption until 2027 is a pragmatic response. The evidence does not support that view.

The new fabrication capacity currently being brought online globally is almost exclusively directed at advanced AI chips and high-margin memory for AI applications. TSMC, which controls over 90% of leading-edge semiconductor production, reported a 58% rise in Q1 2026 profits and announced capital expenditure plans of US$52 to 56 billion for 2026 alone. The overwhelming majority is directed at advanced node capacity for AI customers including NVIDIA, Meta, and Amazon. While this investment will eventually increase total semiconductor output, the near-term supply of conventional enterprise memory will remain constrained because the fabrication lines producing it have been reallocated, not merely paused.

IDC expects memory supply challenges to persist throughout 2026 and well into 2027. Analyst commentary across multiple sources suggests that a meaningful return to 2025 pricing levels is unlikely before 2028, at the earliest.

For organisations considering deferral, this translates to a specific set of risks.

• Prices for new hardware are more likely to remain elevated or increase further in 2027 than to revert to pre-shortage levels.
• Lead times show no structural evidence of improvement before at least Q3 or Q4 2027, meaning deferred projects could face the same or longer delivery timelines.
• Ageing hardware retained beyond its optimal refresh window accumulates technical debt, increases maintenance costs, and may fall outside vendor support contracts, creating security and compliance exposure.
• Gartner has warned that extending device lifetimes by 15 to 20% may solve a short-term budget problem whilst creating medium-term management and security challenges that outweigh the savings.
• Organisations that delay and then attempt to procure simultaneously in 2027 may face a demand surge as deferred refresh cycles collide, further compressing availability.

A different path: the case for the secondary market and global brokerage

Against this backdrop, a growing number of IT procurement teams are revisiting an approach that, until recently, sat at the periphery of mainstream enterprise purchasing: the secondary and refurbished hardware market, sourced through specialist global brokers with deep access to pre-owned and refurbished enterprise-grade equipment.

This is not a new market. It is, however, a market that has matured considerably and which is now attracting renewed attention from CIOs and CTOs who recognise that the constraints they are experiencing with new hardware simply do not apply in the same way to carefully sourced, tested refurbished systems.

The global refurbished IT hardware market was valued at approximately US$10.5 billion in 2026 and is forecast to grow at a compound annual rate of nearly 10% through to 2031. That trajectory reflects the structural shift now underway in enterprise procurement thinking.

The supply dynamics are also more favourable. Secondary market inventory is not subject to the same semiconductor fabrication constraints as new-build systems. Refurbished servers, storage arrays, and networking equipment can frequently be sourced, tested, and delivered in timescales that are a fraction of the 32 to 40-plus week lead times now quoted by major vendors for new hardware.

How Covenco can help: global reach, local expertise

Covenco has operated at the intersection of enterprise IT supply and technical expertise since 1989. As an IBM Gold Business Partner and specialist in managed IT services, Disaster Recovery, and Backup-as-a-Service, we understand the infrastructure decisions that underpin business resilience, and the procurement pressures that complicate them.

Within the current supply chain environment, our position within global broker and secondary-market networks provides something that standard channel procurement cannot: access to a wider, more flexible inventory of enterprise hardware at a time when the primary supply chain is operating under severe constraint.

Covenco sources and supplies a comprehensive range of enterprise IT hardware, including:

• IBM Power systems, from individual components and upgrade parts through to complete, tested server systems.
• NetApp, Pure Storage, and Dell EMC storage arrays, controllers, shelves, and drives.
• Cisco networking equipment, including switches, routers, firewalls, and optical components.
• HPE and Dell server platforms across current and recent-generation architectures.
• Individual components including memory modules, CPUs, HBAs, NICs, SSDs, HDDs, and power supplies for all major platforms.

This capability is supported by warehouse and distribution facilities in the United Kingdom, Ireland, Denmark, Spain, and the United States, enabling Covenco to source from global broker markets and deliver reliably to customers across Europe and beyond. That global footprint matters in a disrupted market: it provides access to inventory pools that are unavailable to organisations relying solely on their incumbent vendor relationships.

Critically, our capability extends beyond the latest generation of hardware. For organisations that have made the considered decision to sustain and extend their existing IT assets rather than commit to new hardware at current prices, Covenco is well placed to supply parts for older systems that are no longer actively stocked by manufacturers or mainstream distributors. Whether that means a memory upgrade for a legacy IBM Power9, a replacement controller for an older NetApp FAS system, or Cisco components for a switching infrastructure that pre-dates the current generation, we can source and supply the parts that keep existing infrastructure operational, performant, and supportable, without requiring an immediate capital commitment to new hardware.

Sustaining existing infrastructure is not a compromise. In the current market, it can be the most commercially and technically rational choice available.

Keeping end-of-life systems compliant and fully supported

Hardware procurement is only part of the challenge for organisations choosing to extend the operational life of existing infrastructure. When a vendor declares a system End of Life (EOL) or End of Service Life (EOSL), the commercial support contract that has underpinned that asset, the firmware patches, the hardware break-fix cover, the escalation paths, typically expires with it.

For many organisations, that creates an immediate compliance risk. Regulatory frameworks including ISO 27001, PCI DSS, SOC 2, and sector-specific mandates such as those governed by the FCA or NHS Digital require demonstrable, documented support arrangements for systems that process or store sensitive data. A lapsed vendor support contract does not automatically mean a system is insecure, but it does mean that the organisation must evidence an alternative support arrangement to satisfy auditors and maintain its compliance posture.

Covenco addresses this directly through third-party maintenance (TPM) contracts that provide structured, documented hardware support for systems that have reached, or are approaching, vendor EOL status. These contracts deliver the break-fix coverage, response SLAs, and technical escalation paths that organisations need to maintain audit-ready compliance, without the cost premium of remaining on an OEM support agreement for ageing hardware.

For organisations running IBM Power, Cisco, NetApp, HPE ProLiant, Dell PowerEdge, or Pure Storage systems on extended lifecycles, a Covenco TPM contract provides the formal support layer that transforms a pragmatic operational decision into a fully defensible compliance position.

An organisation that chooses to sustain a stable IBM Power9 environment for another two years rather than invest in new hardware at current pricing does not have to accept a gap in its support coverage to do so. With the right TPM contract in place, that system retains documented SLA-backed support, change-control-compatible maintenance, and the evidence trail that compliance teams, auditors, and insurers require.

In a market where new hardware is constrained, expensive, and slow to arrive, the ability to extend the supported life of existing systems, compliantly and with full documentation, gives IT teams the time to plan infrastructure refreshes properly, procure at the right moment rather than under duress, and avoid the risks that come with rushing complex platform migrations to meet an arbitrary vendor deadline.

Gaining procurement agility in a constrained market

The current enterprise hardware supply chain crisis is neither temporary nor simple. It is the result of a fundamental reallocation of global semiconductor capacity towards AI infrastructure, a shift that is structural, deliberate, and unlikely to reverse within a timeframe that benefits organisations procuring hardware today.

Gartner, Deloitte, IDC, and leading vendors have all provided consistent signals: prices will remain elevated, lead times will remain extended, and meaningful relief is unlikely before 2028.

For organisations navigating this environment, procurement agility is no longer a ‘nice to have.’ It is a competitive advantage. Organisations that can source the hardware they need, new or refurbished, current-generation or legacy, from a partner with global reach and deep secondary-market expertise, will refresh their infrastructure on their terms, on their timelines, and at pricing that reflects market reality rather than tier-one vendor allocation queues.

Covenco has been doing exactly this for over three decades. If you are facing extended lead times, budget pressure on new hardware, or the challenge of sustaining legacy systems through a period of supply constraint, we would welcome the conversation.