If your organisation has felt the sting of higher prices or longer lead times on servers, storage, memory or end-user devices over the past year, you are not imagining it. The AI build-out is reshaping the global hardware market in ways that go well beyond a short-term price spike.
The key point for IT and business leaders is this: what began as a temporary shock now looks more like a multi-year supply and pricing cycle. That means waiting for the market to normalise is unlikely to be an effective strategy.
At 11:11 Systems, we are having this conversation with customers and prospects every week. Based on what the market is signalling, this is what has already happened, how long the pressure could last, and what organisations should do next.
What is driving the squeeze?
The root cause is straightforward. AI data centres are consuming an enormous share of global memory production.
To support AI accelerators, the suppliers that produce most of the world’s DRAM are redirecting wafer capacity towards high-margin, high-bandwidth memory and server-grade DRAM and NAND. Every wafer used for AI memory is a wafer that does not go into desktop RAM, data-centre SSDs or other components enterprises rely on every day.
Because suppliers are locking in large, long-term contracts with hyperscale AI customers, enterprise and consumer buyers are absorbing more of the volatility. And because new fabrication capacity takes years to bring online, supply cannot respond quickly enough to the surge in demand.
The result is tighter availability and higher prices across a growing range of hardware categories.
What has already happened to prices?
Recent market moves have been steep.
- Conventional DRAM contract prices rose an estimated 58 to 63 percent quarter on quarter in Q2 2026, while NAND Flash contract prices rose 70 to 75 percent, according to TrendForce data cited in external reporting from 36Kr and echoed across the market.
- Enterprise SSD contract prices climbed by roughly 80 per cent in Q1 2026 as AI infrastructure demand consumed available NAND output, according to Astute Group.
- Gartner has been widely cited forecasting that combined DRAM and SSD costs could rise by about 130 per cent by the end of 2026, feeding through higher PC and smartphone prices, including in reporting from CNN.
- Other reporting notes that some memory quotes have only remained valid for a week, making procurement timing far more difficult for IT teams.
There is some good news in the data: the pace of price increases appears to be slowing. Tom’s Hardware reports that TrendForce projects Q3 2026 DRAM contract prices rising by 13 to 18 percent quarter on quarter, which is still significant but materially lower than the jumps seen in Q2.
That said, the slowdown appears to be driven more by affordability limits than by a true improvement in supply.
How long could AI hardware shortages last?
This is the question every leader is asking, and the honest answer is that there is a range of plausible scenarios rather than a single date.
The more conservative view is that pressure continues through 2027 and begins easing in 2028 as new capacity starts to come online. CNN reported in July 2026 that the shortage is expected to persist until at least 2028, while Tom’s Hardware noted that Samsung and SK hynix have warned AI-driven memory shortages could last until 2027 and beyond.
A more extended scenario pushes meaningful normalisation into 2029. Design News cited Kearney saying the earliest meaningful new capacity may arrive in 2028, with fabs ramping through 2029 and no substantial relief before early 2030 if AI demand continues at its current pace. AI hardware shortages persist Rock Paper Shotgun echoed the view that meaningful change in DRAM pricing may not arrive until 2028 at the earliest.
Even if the acute shortage eases, many analysts expect pricing to settle above pre-2026 levels rather than returning to the old baseline. For most organisations, that means planning for a structurally higher cost floor rather than a rapid reset.
Two variables could still change the timetable in either direction:
- additional manufacturing capacity entering the market faster than expected
- improvements in AI model memory efficiency that soften demand growth
Neither outcome is guaranteed.
What this means for your organisation
The practical implication is simple: hoping the market resets is not a strategy.
IT and business leaders should consider the following actions now:
- Plan earlier. Bring forward hardware refreshes, migrations, resilience initiatives, storage growth decisions and AI infrastructure planning where it makes sense.
- Secure supply and pricing windows. Where contracts allow, lock in earlier and expect shorter quote validity periods.
- Do not budget for a rapid reset. Assume elevated pricing persists through 2027, with only gradual easing across 2028 and potentially 2029.
- Prioritise by risk. Focus first on the workloads and user groups where hardware constraints would create the greatest business, resilience or security exposure.
- Reduce exposure to volatile hardware cycles. Lifecycle extension, phased roll-outs, managed services and cloud consumption models can all reduce dependence on spot-market component pricing.
How 11:11 Systems helps reduce exposure
This is exactly the kind of market environment where a resilient, service-led approach can pay off.
11:11 Systems helps organisations reduce direct dependence on volatile hardware procurement cycles through 11:11 Hosted Private Cloud, the broader 11:11 Cloud platform, the 11:11 Cyber Recovery Platform, and related managed infrastructure services. Instead of tying every major initiative to hardware purchasing windows, organisations can shift more of that risk into a managed consumption model.
Our advisory and consulting teams also help leaders prioritise, sequence and de-risk major infrastructure decisions, so they can make deliberate choices rather than reacting to the next price increase or delivery delay. If you are assessing how cloud strategy affects infrastructure flexibility, our recent article on VCF 9, Infrastructure, and the AI Revolution offers a useful companion read.
The AI hardware shortage cycle is likely to stay with us for several years. The organisations that plan for it deliberately, rather than hoping for a quick return to old pricing, will be better positioned to stay resilient and in control of costs.
If you are reviewing a refresh, resilience or infrastructure roadmap, now is the right time to reassess how exposed you are to hardware volatility.
Additional resources
- VCF 9, Infrastructure, and the AI Revolution
- Unlock AI with GPU as a Service in VCF 9
- Supply Chain Whiplash: Why Your Orders Keep Slipping
- How to Bring Predictability to Tech Supply Chain Disruptions
- Why 11:11 Systems
