Rising memory costs already affecting most IT buyers
Vespertec research reveals widespread impact on infrastructure planning, with organisations extending hardware lifecycles and delaying projects as constraints persist.
More than half (53%) of IT decision-makers are already experiencing the impact of rising memory costs, with a further 39% expecting to be affected soon, as sustained pressure driven by AI workloads continues pushing demand beyond what the global infrastructure supply can absorb, according to new research from data centre infrastructure specialists, Vespertec.
The study, based on responses from 305 IT decision-makers who are involved in hardware purchase decisions, maps a sustained challenge that is now shaping costs, planning timelines, and delivery across organisations.
Rising costs and shifting demand are creating new pressures
Over a third (37%) of respondents say their biggest concern is the difficulty of justifying higher costs internally, which significantly outweighs concerns about securing supply (12%).
The research also suggests that demand dynamics are changing. Nearly two-thirds (64%) of respondents identify AI-driven demand and hyperscaler consumption as the primary cause of the shortage, ahead of geopolitical disruption (13%).
Organisations are adapting through short-term, pragmatic measures
In response, organisations are prioritising practical, near-term adjustments as they manage constraints.
The impact is already being felt at a project level. More than half of organisations are experiencing disruption today, and a further 39% expect to be affected soon. Among those already impacted, nearly three-quarters (72%) report having already delayed or cancelled projects planned for 2026. What were once contingency measures are now becoming embedded in day-to-day infrastructure decision-making.
Among those already impacted by DRAM price increases, the most common measures include extending the lifespan of existing hardware (49%), delaying or rescoping infrastructure projects (37%), and redesigning systems to reduce memory dependency (31%). Alongside this, many organisations are rebalancing the trade-off between cost and performance, with nearly one in five (18%) accepting reduced performance in order to remain within budget.
The research also uncovered signs of changing supplier dynamics. Among organisations already affected, over half (53%) of organisations are consolidating or considering consolidating their supplier base. Organisations are prioritising reliability and guaranteed access over maintaining a broad range of vendors, pointing to a shift toward closer, more strategic supplier relationships in response to ongoing constraints.
Limited visibility and ongoing constraints are shaping long-term outlook
Looking ahead, most organisations do not expect conditions to ease quickly. Eight in ten respondents (84%) anticipate the shortage will persist for at least 12-18 months, and around 90% expect delays to infrastructure projects over that period.
Simultaneously, a lack of clear information is making it harder to plan effectively. Among those already impacted, only 15% of organisations say they receive clear, actionable insight from vendors, leaving many to navigate uncertainty around pricing, availability, and lead times.
“What we’re seeing in the market right now is very real pressure on memory supply. Demand is rising across the board, particularly from AI, and manufacturers are having to make difficult trade-offs,” said Allan Kaye, Co-Founder of Vesper Technologies. “In many cases, production is being prioritised towards HBM3 to support AI platforms, which is further constraining the supply of DRAM used more broadly across server infrastructure.
“Recent reports of server vendors struggling to fulfil certain configurations show how tight parts of the market have become. For most organisations, however, cost is emerging as the more immediate challenge. As prices rise, many teams are under increasing pressure to justify spend internally, which is now a more significant barrier than securing supply itself.
“AI platforms still rely heavily on DRAM, so the impact of these supply constraints is being felt across the wider market. At the same time, new capacity takes time to come online, and demand isn’t slowing. That leaves organisations in a difficult position – waiting for prices to stabilise carries risk, but reacting too quickly can be just as challenging.
“The organisations navigating this most effectively are planning further out. They’re reassessing priorities, what they actually need, working closely with partners to secure the infrastructure they’ll need over the next 12 to 24 months.”
