TL;DR
A late-June 2026 analysis from Thorsten Meyer AI says memory and storage costs have become the main price shock for high-end PC and workstation buyers. HP told investors memory rose from 15% to 18% of PC bill-of-materials costs to about 35%, making DIY builders more exposed to retail price swings.
High-end PC builders and workstation buyers are facing a new cost shock as memory and storage prices become one of the largest parts of a system build, according to a late-June 2026 report from Thorsten Meyer AI.
The report says RAM and SSDs have moved from secondary expenses to a central cost in premium machines. It cites HP as telling investors that memory rose from 15% to 18% of a PC bill of materials to about 35% in a single quarter.
Thorsten Meyer AI said one 2026 build comparison showed a 32GB DDR5 kit priced at about $369, roughly in line with the graphics card used in the same build. The report says premium builds that cost about $2,000 a year earlier now fall in the $2,800 to $4,500 range, with memory and storage driving much of the increase.
The report also says the old assumption that DIY builds reliably save money has weakened at the high end. Large PC makers and system integrators can buy memory through bulk contracts or draw from earlier inventory, while individual builders usually pay the retail spot price at the time of purchase.
The high-end PC & workstation tax
If you build your own machines or spec your team’s workstations, you’re the most exposed buyer in this market — no hedge, no bulk contract, just a parts cart and a number you used to ignore, now the biggest line on the invoice.
OEMs buy on bulk contracts and hold hedged stock; you pay the spot price on the day. The DIY builder is now the most exposed buyer in the chain — and the prebuilt is sometimes cheaper. Price it before you commit.
96GB & 128GB DDR5 RDIMMs are the scarcest, closest to the server memory makers prioritize. 64GB RDIMM could cost 2× by end-2026 vs early 2025. The parts that define a workstation are the ones squeezed hardest.
The squeeze didn’t just raise prices — it inverted the value system of high-end building. Buy big, buy early, build it yourself: each enthusiast virtue is now a way to overpay. Discipline beats ambition in 2026 — right-size hard, buy deliberately, lean on bundles, treat the prebuilt as a real price check. You can’t avoid the AI tax levied a layer up in the fabs; you can refuse to pay more of it than the job needs. Next: Cloud’s Hidden Memory Bill.
Retail Buyers Bear The Shock
The impact is most direct for people who build systems themselves or specify machines for small teams. A buyer choosing parts one at a time has less protection from sudden component price moves than an OEM with contract pricing, inventory planning, and supplier leverage.
That changes the value calculation for enthusiast PCs, creator rigs, and technical workstations. Building still offers control over parts, repairability, and upgrade paths, but the report says it no longer guarantees the lowest price when memory costs are moving quickly.
32GB DDR5 RAM kit
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
AI Demand Hits The Workbench
The report frames the issue as part of a broader 2026 memory crunch tied to demand for high-bandwidth and server memory. Earlier parts of the series traced pressure from HBM through standard RAM and storage before focusing on end users buying complete machines or individual parts.
The workstation market is described as especially exposed because many professional systems need 64GB, 128GB, or more of memory. The report says 96GB and 128GB DDR5 RDIMMs are among the tightest categories because they are close to the server-grade parts that memory makers are prioritizing.
“Memory went from afterthought to the biggest line item.”
— Thorsten Meyer AI report
high-end SSD NVMe 1TB
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Price Path Still Unsettled
It is not yet clear how long the pricing pressure will last or how far retail memory prices will move through the rest of 2026. The report says its cited prices are point-in-time figures from late June and are changing quickly.
Forecasts for workstation memory remain claims rather than settled outcomes. The report cites analysis projecting that 64GB DDR5 RDIMMs could cost about twice as much by the end of 2026 compared with early 2025, but future pricing will depend on supply, AI infrastructure demand, and manufacturer allocation decisions.
gaming PC build components
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Builders Reprice Their Parts Lists
The report recommends that buyers right-size memory needs, compare prebuilt systems before ordering parts, use CPU or motherboard bundles where available, and delay upgrades that are not immediately needed.
For workstation buyers, the next step is likely closer scrutiny of capacity requirements. A team that once ordered 128GB as a safety margin may now have a reason to separate immediate workload needs from later expansion.
professional workstation parts
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is the high-end PC and workstation tax?
It refers to the extra cost buyers now face because RAM and SSD prices have risen sharply, making high-capacity builds more expensive than expected.
Does building a PC still save money in 2026?
Not always at the high end, according to the report. DIY builds still offer control, but prebuilt systems may be cheaper in some cases because OEMs can use bulk pricing or existing inventory.
Why are workstations hit harder than gaming PCs?
Many workstations need large RDIMM capacities, such as 96GB or 128GB modules. Those parts overlap more closely with server memory demand, where supply is tight.
What should buyers do before ordering parts?
The report advises checking a comparable prebuilt price, avoiding unnecessary memory overbuying, using bundles where they reduce cost, and staging upgrades when possible.
Source: Thorsten Meyer AI
