Goldman sees “Adjustment Storm” in DRAM: the spot surges while the contract lags

The memory market is once again at the center of the tech landscape, not only because of the AI craze. A new warning from Goldman Sachs —reported by Asian financial press— points to an unusual imbalance between the spot price (immediate purchase) and the contract price (forward agreements) in DRAM. The message is clear: when this “scissors” opens too wide, it typically closes in one way or another. In previous cycles, the most common way to “close” it has been a sharp rise in contract prices.

The troubling figure: DDR4 +172% on contracts

The data fueling headlines is striking: DDR4 is showing a spot premium of up to 172% over the contract; for DDR5, the premium hovers around 76%. In other words: the “cash” market is already paying much more for modules today, while contracts still don’t fully reflect that tension.

Practically, this often triggers a domino effect. If manufacturers and distributors see that the spot channel “validates” higher prices (due to scarcity or urgency), quarterly or semi-annual contract negotiations become more aggressive, because buyers lose leverage to push prices down.

Why does this “scissors” happen: spot vs. contract don’t react the same

In memory, the spot and contract markets operate on different rhythms:

  • Spot: reacts quickly to bottlenecks (low inventories, demand spikes, logistical delays).
  • Contract: moves more slowly, as it depends on periodic renegotiations, committed volumes, and supply relationships.

When the spot price diverges too much, the market is essentially signaling: there’s real supply/demand tension, and someone is paying for priority.

AI drives demand… but also reshapes supply

The 2026 landscape is especially delicate because the memory industry is rebalancing toward higher-value products (HBM for AI accelerators) and, simultaneously, experiencing increased demand for “conventional” DRAM for servers, PCs, and devices.

In fact, TrendForce describes a “vendor’s market” scenario where major manufacturers are pushing for shorter contracts and more frequent revisions, anticipating phased price increases. There’s also talk of significant rises in server DRAM at the start of 2026.

Early signs of price hikes on the industry radar

Beyond the spot vs. contract debate, one sees signs of tightening prices and availability:

  • TrendForce indicates that conventional DRAM contract prices could increase by 55–60% quarterly in Q1 2026, while NAND may rise 33–38% quarterly.
  • In servers, the pressure is even more evident: the sector assumes that the gap between supply and demand is widening as hyperscalers and large buyers secure capacity ahead of the rest of the market.

This aligns with the core diagnosis: AI is not only driving up compute chips but also lifting memory (HBM and DRAM), infrastructure, inventories, and industrial planning.

Implications for companies and consumers

For companies (IT and procurement)
If contract correction happens forcefully, it will impact budgets for server upgrades, RAM expansions, and capacity projects. It’s not just about paying more: rapid price hikes often come with tighter timelines, early purchases, and pressure to make decisions sooner.

For end-users (PCs, laptops, mobile devices)
When DRAM becomes more expensive, the effect usually filters into product prices and, in some cases, less generous configurations (base models with less RAM or big jumps between variants). In mobile devices, if NAND also gets more costly, the impact will be felt in storage options and entry-level segments.

What to watch now

In the coming weeks, focus will be on two fronts:

  1. Contract renegotiations in the first half of 2026: this will reveal if the “adjustment” suggested by Goldman Sachs materializes.
  2. Actual capacity (and how it’s allocated): if HBM continues to attract industrial priority, conventional DRAM may become more constrained, increasing the risk of new spikes.

In summary: if the spot price is so far ahead of the contract, the market is warning that memory is once again a strategic bottleneck. And in the race for AI, bottlenecks rarely resolve without passing through the checkout.


Frequently Asked Questions

What does it mean that DDR4 has a spot price 172% higher than the contract?
It means that for immediate purchases, the market is paying significantly more than the agreed-upon forward prices, a typical sign of scarcity or urgency.

Why does AI also increase the cost of “regular” RAM?
Because the AI boom drives demand for servers and accelerates production of advanced memories (like HBM), which can strain supply chains and available capacity for conventional DRAM.

Could this affect laptop and PC prices in 2026?
It’s possible, especially if DRAM contracts “catch up” with the spot prices and manufacturers pass some of the cost increases to the final price or cut down on base configurations.

How long might such an upward cycle last?
Memory cycles are recurring: these episodes can last several quarters if demand sustains and supply doesn’t respond with enough capacity or inventory. In 2026, the key variable is the sustained pressure from AI on the entire supply chain.

Scroll to Top