If you sell technology into the channel, your inbox right now probably looks a lot like ours. Price-change notices from upstream suppliers. Customer emails asking why last month's quote is no longer honorable. Purchasing managers requesting allocation commitments you can't make. The memory market has entered its most disruptive pricing cycle in more than a decade, and the explanation most resellers are giving customers is not quite landing.
This post is the talk track. It covers what is actually happening, why it is happening, how long it is likely to last, and - most importantly - what to tell your customers so they can make good decisions in the middle of it.
What is actually happening with memory prices
According to wccftech.com 2026 Q1 Memory prices increased 110%. NAND flash contract prices arose 147% over the same period. Server DRAM specifically is up roughly 90 percent QoQ, marking the largest single-quarter increase on record. PC DRAM is expected to at least double.
Those are contract prices, not spot prices. The distinction matters. Contract prices are what OEMs and large buyers negotiate for multi-month supply. Spot prices, which fluctuate daily, have been running even higher as buyers without allocations scramble for inventory. When you hear that memory is up 40 percent, 60 percent, or 100 percent, the answer depends on which price someone is quoting.
For practical purposes, every category of memory and storage component is more expensive in Q1 and Q2 2026 than it was six months ago, and the direction of travel is still up.
Why this is happening
Three forces are colliding at once, and none of them are going away quickly.
1. AI data centers are absorbing premium capacity
Hyperscale cloud service providers - the North American operators building out GPU-heavy AI infrastructure - have been locking in long-term agreements with memory suppliers since late 2025. Industry analysts estimate AI data centers will consume roughly 70 percent of high-end DRAM in 2026. High-bandwidth memory, the stacked DRAM used on modern GPUs, is even more heavily allocated; leading vendors have indicated HBM capacity is booked through 2026 and into 2027.
When suppliers commit capacity to HBM and high-density server DDR5, that capacity is not available for the PC, workstation, and commercial desktop/laptop market your customers are buying for. This is the single biggest driver of the squeeze.
2. Suppliers are reallocating wafer capacity
Memory manufacturers - Samsung, SK hynix, and Micron being the three largest — are deliberately shifting production from NAND flash to DRAM, and within DRAM from conventional modules toward server and HBM. They can do this because the margins on AI-class memory are dramatically higher than on commodity modules. The side effect is that NAND production capacity is growing more slowly than NAND demand, which is why SSD pricing is also climbing.
3. DDR4 is being phased out
Industry plans have DDR4 entering end-of-life through late 2025 and early 2026. Counterintuitively, EOL transitions often push prices up rather than down in the short term: suppliers cut production runs before demand catches up with the next generation, and customers with DDR4-only platforms have to buy against a shrinking supply pool. This matters for anyone servicing customers whose installed base has not yet refreshed to DDR5-capable platforms.
What to tell your customers
Three conversations are worth having with every customer who asks about pricing.
Conversation 1: Lead times matter more than unit cost
In a normal market, the negotiation is about price. In this market, the real question is whether the inventory will be available at all when the project needs it. Encourage customers to prioritize suppliers who can commit to delivery windows in writing and back those commitments with inventory. A part that ships on schedule at a higher price is cheaper than a part that misses the deployment window entirely.
Conversation 2: Quote validity is shrinking
Standard 30-day quotes do not work in this environment. Several large OEMs, including Dell, have publicly acknowledged shorter quote windows and mid-cycle repricing on memory-heavy configurations. Be direct with customers: quotes need to be converted to purchase orders within days, not weeks. If the customer can commit now, lock in the price. If they cannot, make sure they understand that requoting in two weeks will produce a different number.
Conversation 3: Pull projects forward where possible
For any customer with a Q3 or Q4 2026 refresh project, the conversation is about whether it makes sense to pull that project into Q2. Most analysts expect the supply-demand gap to persist into 2027, and contract prices are forecast to keep climbing through the first half of 2026. If budget and deployment logistics allow, buying ahead of the curve is cheaper than buying on it.
How Can We Help
Bull Creek Technologies operates on a factory-direct sourcing model, which means our memory and SSD products come from qualified Tier 1 manufacturing partners without the brand-layer markup. In a normal market, that translates to better reseller margin. In a constrained market, it also translates to cleaner supply signals and more predictable allocation. We cannot promise immunity from an industry-wide shortage — nobody can — but we can promise direct visibility into lead times and pricing, without a brand's pricing team setting policy in between.
For now, the guidance is straightforward: buy what you can, lock in what you quote, and keep customers informed. If you want to talk through specific customer situations or get current lead-time and pricing intelligence on BCT memory and SSD products contact us at info@bullcreektech.com