Tech headlines are being dominated by the perfect storm that has led to a global shortage of Random Access Memory (RAM).
As the short-term, temporary memory that handles data for processing and applications, RAM – and specifically Dynamic Random Access Memory (DRAM) – is a foundational business technology.
The primary driver of this shortage is an industry-wide shift to High-Bandwidth Memory (HBM). This is the specialised memory required for artificial intelligence (AI). In the face of voracious AI demand, major manufacturers have reallocated up to 25% of their production capacity to meet this AI-specific market.
This has effectively strangled the production – and consequent supply – of standard DRAM used in servers, PCs, and networking gear.
But for those in digital infrastructure, this has not been the only impact of the intense demand for AI hardware. Hyperscale providers such as Microsoft, Google and Meta have used their massive capital reserves to secure long-term supply contracts for this HBM.
In effect, they have bought the future and ‘guaranteed’ the HBM market. This has turned a short-term inflation of demand into a new normal, leaving everyone else in the mid-market, and smaller businesses, fighting for the remaining inventory of DRAM.
Not only has this led to extreme price hikes – memory prices have surged approximately 90% in Q1 2026 according to Counterpoint Technology Market Research – it has extended lead times massively. Large DRAM orders now frequently exceed 40 weeks.
This has already caused significant delays. Infrastructure projects slated for 2026 have been pushed into 2027 or 2028.
In the interim, businesses have been denied the opportunity to simply sweat existing assets, as legacy technology is being phased out faster than expected. For industrial, medical, and automotive sectors that rely on older hardware, this is now a critical exposure to risk.
Businesses that rely on a small group of suppliers and producers are feeling the biggest impact. Customers now face profound uncertainty as the pricing they need to factor in for digital infrastructure projects is dramatically skewed.
Many analysts predict this situation will run until mid 2027 at the earliest. Businesses that cannot lock in 18 months of requirements now, risk being discarded by vendors in favour of higher-margin AI customers.
Many businesses face either having to cancel their digital infrastructure plans, or make fundamental changes in how those projects will be realised.
Infrastructure as a Service (IaaS) is one viable answer. That’s because terabytes of RAM are readily available, offering an effective alternative to weather the storm. Immediate capacity is already in place, and some providers, such as Pulsant, have even secured access to enough memory to service growth targets.
This IaaS shield against volatility does not just negate the 40-week delays and 90% price hikes. It means that instead of a business over-purchasing RAM at peak prices to "future-proof" themselves, they can instead pay for the memory they need today, without risking their business continuity.
Put simply, that means money is saved and the business is protected.
This current, acute crisis stems from deliberate capacity reallocation as markets have changed. The major question facing businesses is if the direction of that change will reverse.
There are signs that no such ‘return to memory as a commodity’ will happen any time soon. Samsung Electronics has announced plans for 50% HBM capacity surge in 2026, whilst SK Hynix has already begun HBM4 mass production.
However, analysts suggest that if additional memory production facilities come online as planned, late 2027 supply chains may normalise; especially if AI demand also eases.
But, for mid-market and smaller businesses, survival and growth in the interim is not guaranteed. Welcome to the year of living dangerously.
