RAM price crisis extends through 2026 as GPUs follow suit

Craig Nash
By
Craig Nash
AI-powered tech writer covering artificial intelligence, chips, and computing.
7 Min Read
RAM price crisis extends through 2026 as GPUs follow suit — AI-generated illustration

The RAM price crisis shows no signs of abating. Framework, the modular laptop maker, issued a stark warning in its latest pricing update: customers should prepare for continued volatility and cost increases throughout the remainder of 2026. The company’s blunt assessment comes as GPU prices suddenly spike alongside memory costs, creating a compounding headache for anyone building or upgrading a PC.

Key Takeaways

  • Framework warns the RAM price crisis will persist with volatility and increases through late 2026
  • LPDDR5X memory surged recently, forcing Framework to raise Desktop prices while absorbing some costs
  • Storage Expansion Cards face substantial price hikes in 2026 due to rising NAND flash costs
  • GPU prices are climbing due to higher VRAM pricing alongside system RAM shortages
  • OpenAI’s 2025 supplier deals securing ~40% of global RAM production for AI data centers triggered the crisis

What Triggered the RAM Price Crisis

The RAM price crisis erupted after OpenAI secured exclusive deals with Samsung and SK Hynix in 2025, locking down roughly 40% of global RAM production for AI data center operations. This massive diversion of supply away from consumer and commercial markets created an artificial scarcity that sent prices skyrocketing. DDR5 memory kits have climbed approximately 120% from baseline prices, with some kits jumping from around $100 to substantially higher levels. DDR4 32GB kits now cost $120 to $150, up from previous lows, while 16GB kits range from $75 to $100.

Framework’s own pricing reflects this chaos. The company raised memory module costs for the second time in a month around December 2025 and January 2026, citing extreme shortages. Today, 32GB modules cost roughly $10 per gigabyte, while 48GB modules hit approximately $13 per gigabyte. The company expects another price increase by January 2026 and warns of continued turbulence through the year. One analyst reviewing the situation noted that the magnitude of price increases seems disconnected from supply-and-demand fundamentals alone.

Framework’s Response and Customer Options

Rather than absorb all costs, Framework is encouraging customers to source their own RAM through PCPartPicker links integrated into its configurator for DIY Editions. The company claims this approach offers pricing favorable compared to Apple’s roughly $25 per gigabyte for memory upgrades. Framework is also absorbing a portion of LPDDR5X cost increases for its Desktop configuration with 128GB of memory, though the exact price increase was not disclosed. The company pledges to stay as close as possible to acquisition costs, a commitment tested daily by volatile supplier pricing.

For those building systems with Framework’s Storage Expansion Cards, expect more pain ahead. The 250GB version has already seen a price increase, and both the 250GB and 1TB variants will likely face substantial adjustments throughout 2026 due to rising NAND flash costs. Framework noted some signs of stability in March, but cautioned against viewing this as anything more than a temporary reprieve.

GPUs Join the Price Hike Party

The RAM price crisis is now spilling into GPU pricing. Graphics card costs are rising due to higher VRAM pricing alongside the broader system RAM shortage. This creates a cascading problem: anyone upgrading a PC faces higher memory costs, higher storage costs, and now higher GPU costs simultaneously. The root cause remains the same—AI data center demand is consuming supply faster than manufacturers can replenish it, and the industry has no clear timeline for relief.

Analysts predict the supplier cost trend will persist through 2026, with some price stabilization or plateau possible in late 2025 or early 2026, but supply tightening continuing. This means the RAM price crisis will remain a fact of life for PC builders, laptop upgraders, and anyone with a system refresh on their roadmap. Framework’s warning is not alarmist—it is a realistic assessment of a market fundamentally distorted by AI infrastructure investment.

Should You Wait for Prices to Drop?

Waiting for memory prices to return to 2024 levels is likely a losing strategy. Framework’s explicit warning of volatility through the end of 2026 suggests the market will not stabilize in any meaningful way within the next year. If you need a system upgrade or a new laptop, delaying further compounds the problem—prices are more likely to rise than fall in the near term.

Is the RAM price crisis affecting all memory types equally?

No. DDR5 has seen roughly 120% price increases, while DDR4 has climbed more modestly but still significantly. LPDDR5X, the memory used in Framework’s high-end Desktop configurations, surged recently enough to trigger a price hike. The crisis affects all consumer memory segments, but the magnitude varies by type and capacity.

What can I do to avoid high memory prices?

Framework’s strategy of sourcing your own RAM via PCPartPicker offers one path, particularly if you are comfortable building or upgrading a system yourself. Comparing prices across vendors and considering older DDR4 systems might offer modest savings versus DDR5, though the premium for newer memory remains substantial. For laptops with soldered memory, you have no choice but to pay what manufacturers charge.

The RAM price crisis is not a temporary blip—it is a structural problem created by AI infrastructure capturing global memory supply. Framework’s warning, backed by concrete price increases and inventory constraints, signals that 2026 will remain expensive for anyone upgrading memory, storage, or GPUs. Plan your builds accordingly and budget for higher component costs than you would have expected two years ago.

This article was written with AI assistance and editorially reviewed.

Source: TechRadar

Share This Article
AI-powered tech writer covering artificial intelligence, chips, and computing.