Why RAM Got So Expensive, and Why It’s Not Over Yet
For years, buying more RAM was one of the easier ways to improve a PC. Prices were generally predictable, capacities kept increasing, and a 16GB or 32GB upgrade was a staple threshold in building or upgrading a computer.
That changed dramatically over the past year.
By late 2025 and into 2026, memory prices began climbing as the semiconductor industry struggled to keep up with demand for D-RAM.
What initially looked like another temporary memory shortage has become a much bigger issue, driven by one of the technology industry’s biggest trends: artificial intelligence.
And if you were hoping RAM prices would have returned to normal by now, the latest outlook is not particularly encouraging.
Table of Contents
The AI Boom Is Eating Into the RAM Supply
The shortage is easier to understand when you look beyond the RAM installed in a desktop PC.
AI data centers require enormous amounts of memory, particularly high-bandwidth memory, or HBM. These specialized memory chips work alongside powerful AI accelerators to process the huge amounts of data modern AI models need.
RAM is only one part of the larger technology stack supporting modern AI systems. To understand how memory, processing, storage, and connectivity work together, see our guide to AI infrastructure and data.
HBM is not the same thing as the DDR4 or DDR5 RAM you buy for a computer. However, both rely on the same broader DRAM manufacturing ecosystem.
Producing HBM also requires considerably more wafer capacity than conventional DRAM, meaning manufacturers have strong incentives to prioritize it when AI companies and data-center operators are willing to pay more.
The people building AI infrastructure aren’t necessarily competing with PC builders for the same RAM modules, but they are competing for a limited manufacturing resource.
The growing demand for AI accelerators is also putting pressure on the underlying compute infrastructure required to run these workloads. For a closer look at the hardware powering modern AI systems, see our guide to AI compute and GPU infrastructure.
It is a little like a GPerya booth suddenly having one game become wildly popular: the booth can only serve so many players at once, so more attention and resources naturally go toward the activity attracting the biggest crowd.
Why Did RAM Prices Rise So Quickly?
Memory manufacturing has always been cyclical. Manufacturers expand production when demand is strong, eventually creating excess supply, after which prices fall and production gets reduced.
The recent situation is different because AI introduced a huge new source of demand, while memory manufacturers tried to protect their margins.
By 2026, major memory companies, including Samsung, SK Hynix, and Micron, were directing more production toward HBM and server-related products. Meanwhile, conventional DRAM remained relatively constrained.
TrendForce estimated that the DRAM market would remain extremely tight through the third quarter of 2026, with conventional DRAM contract prices still expected to rise by 13–18% quarter over quarter.
The result has reached consumers.
PC manufacturers have faced higher component costs, while people building their own computers have seen memory kits become considerably more expensive than they were a year earlier.
The same broader shortage has also affected smartphones, graphics memory, and other electronics that depend on memory chips.
J.P. Morgan’s August 2026 analysis estimated that DRAM prices could rise by more than 400% between the beginning of 2024 and the end of 2026.
Is the RAM Shortage Getting Better in 2026?
Not really.
Some signs point to the memory market beginning to change, but conventional DRAM remains under considerable pressure.
TrendForce expects AI-related demand to continue driving memory consumption into 2027, while DRAM supply remains constrained by continued HBM production and server demand.
The industry is working on increasing supply, but new factories do not appear overnight.
Memory manufacturers are investing heavily in additional production capacity, but semiconductor facilities require years to construct, equip, and bring up to full production.
Deloitte expects meaningful new capacity to take until around 2029 or 2030 in some cases.
One thing’s for sure: there won’t be any haste in returning to the inexpensive RAM prices consumers enjoyed before the shortage.
What Does This Shortage Mean for PC Buyers?
For someone building or upgrading a PC, the situation makes memory capacity a more important purchasing decision.
The obvious response to expensive RAM is to buy less, but that can create another problem.
Modern games, creative applications, browsers, and productivity software can all consume substantial amounts of memory, especially when several applications are running simultaneously.
The better approach is to consider how much RAM you actually need rather than automatically choosing the largest available configuration.
For everyday computing, a sensible amount can still provide plenty of headroom. Users who regularly work with video editing, 3D applications, virtual machines, large datasets, or other memory-intensive workloads may benefit from significantly more.
Memory capacity is only one consideration when designing systems for data-intensive workloads. AI environments also depend heavily on fast, scalable storage and high-speed communication between systems. You can learn more about AI storage infrastructure and AI networking infrastructure to see how these components fit into the wider AI technology stack.
RAM is only one component of a PC. Spending a large portion of a limited upgrade budget on additional memory may make less sense than balancing the system with a faster processor, stronger graphics card, or faster storage.
When Will RAM Prices Finally Come Down?
That is the difficult question.
The memory industry is adding capacity, but AI infrastructure continues to create enormous demand.
TrendForce expects DRAM supply to remain tight through 2027, with more substantial production contributions from new capacity potentially arriving later.
The market may eventually swing too far in the opposite direction.
Memory has historically gone through boom-and-bust cycles, where manufacturers expand production aggressively and later find themselves with more supply than customers need.
However, it is difficult to predict exactly when the balance will shift, particularly while companies continue pouring money into AI infrastructure.
For now, the safest conclusion is that RAM is not returning to its old pricing environment anytime soon.
Until manufacturers can add enough capacity to satisfy both the rapidly expanding AI sector and conventional electronics, consumers should expect memory prices to remain an important part of the cost of building a computer.
