The rapid rise in DRAM prices could severely impact the PC gaming market. You may have already seen the insane price increases of DDR5 memory, with many popular bundles doubling in price within just a few weeks. But rising DRAM costs could also drive up graphics card prices. How bad could the situation get? Should you consider buying a graphics card now to avoid potential price hikes in the coming weeks? That's the question we'll be discussing today.

So why are DRAM prices in this state? In short: it's due to artificial intelligence (AI).

Recent agreements between AI companies and data center builders have set aggressive timelines, aiming to build gigawatt-scale data center capacity within just two to three years. AI data centers require massive amounts of DRAM to meet the memory demands of modern AI models, especially the memory needs of graphics cards.

To support these projects, existing DRAM supply and future capacity have been fully booked. For example, major DRAM manufacturer SK Hynix recently stated that "next year's DRAM, NAND, and HBM (high-bandwidth memory) capacity is completely sold out."

One key player is OpenAI and its "Stargate" project, which aims to rapidly expand server capacity. The two DRAM giants—Samsung and SK Hynix—signed agreements last month to supply memory for this project. Part of the agreement focuses on expanding production, aiming to "achieve a production rate of 900,000 DRAM wafers per month through accelerated capacity expansion."

Reports suggest that the Stargate project will consume 40% of the global DRAM supply, but this isn't entirely accurate. OpenAI's agreement is a plan to expand production and increase capacity, not to immediately consume 900,000 wafers per month. Nevertheless, it's still a massive infrastructure project that will consume a significant share of the memory supply over the next few years.

At this point, you might wonder, what does this have to do with the PC gaming market? Aren't the memory modules used in servers different from traditional DDR5 memory modules? Don't data center graphics cards typically use HBM instead of GDDR memory?

Indeed, but the core issue lies in manufacturing capacity. DRAM factories can adjust production priorities and are capable of producing multiple types of memory. Currently, they are shifting capacity from DDR or GDDR memory to HBM, leading to a reduction in consumer-grade memory supply, which in turn causes price increases and supply difficulties.

Does this sound familiar? This is essentially a replay of the cryptocurrency boom. Due to limited factory capacity and many products being manufactured based on the same process node, manufacturers at that time prioritized the more profitable cryptocurrency market over the gaming market. Today, the same situation is playing out again with the combination of DRAM and AI.

DRAM prices have more than doubled

In the DDR5 desktop memory market, we've seen a significant price impact. Since mid-September, the prices of popular DDR5 kits have more than doubled and haven't yet stabilized, making future price increases uncertain.

Prices for all DDR5 kits have risen sharply, especially the 32GB DDR5-6000 kits we typically recommend. According to PC Part Picker, these kits averaged around $125 for most of the year, but now exceed $250.

The same is true for 64GB kits, which have risen from a low of around $200 in early 2025 to an average of nearly $500 today. Even entry-level DDR5 hasn't been spared: 32GB DDR5-4800 kits, previously priced below $100, have now surged to nearly $200, making slower or cheaper options no longer an option for budget-conscious PC builders.

With DDR5 DRAM module prices continuing to rise month by month, and supply from major manufacturers expected to remain severely constrained throughout 2026, the DDR5 price situation is dire and could worsen further.

This is not the kind of situation where prices will fall or even stabilize after a month or two—spending at least $200 to $250 on memory is rapidly becoming the new minimum standard.

Graphics card prices may be affected

For gamers, the bad news doesn't stop there, as GDDR memory used in graphics cards is also being affected. We haven't seen significant price fluctuations yet, as graphics card prices only stabilized at the Manufacturer's Suggested Retail Price (MSRP) level in early November, but prices are likely to be impacted in the coming months.

This is because GDDR memory shares manufacturing capacity with other types of DRAM, and as manufacturers prioritize AI products, the supply of memory for the consumer market is disrupted.

To illustrate the potential impact on graphics card prices, we conducted research and modeling across different scenarios. First, we need to define the baseline cost of memory in current products. Based on feedback from graphics card manufacturers and information from sources like DRAM Exchange, the cost of current-generation GDDR memory is typically between $2.50 and $3 per GB.

Prices will fluctuate monthly and yearly, but this gives us a general idea of the cost of GDDR6. The price of GDDR7, which powers most Nvidia GeForce 50 series graphics cards, remains unclear, as it is currently an Nvidia exclusive. However, industry sources told us earlier this year that GDDR7 would only be slightly more expensive than GDDR6, not significantly so.

This means that, at a typical benchmark price, the cost for a manufacturer to equip a graphics card with 16GB of GDDR memory is around $40 to $50. GDDR6 typically costs just over $40, while GDDR7 is closer to $50. Graphics cards with 12GB of VRAM (such as the RTX 5070 or Arc B580) will cost slightly less, with GDDR memory costing around $30 to $40. Graphics cards with 8GB of VRAM will cost less than $25.

While this sounds relatively affordable considering that the VRAM cost for most models is less than $50, this basic bill of materials (BOM) cost is not the final price the consumer pays. When graphics card manufacturers like Nvidia and AMD sell GPU cores to their affiliated motherboard manufacturers (AIBs), they bundle the GPU and memory together. This ensures the memory is validated and allows companies like Nvidia to profit by purchasing the same GDDR7 modules in bulk for use by all their AIBs.

Bundling memory with the GPU increases the overall production cost of the product. These companies (especially Nvidia) are known to take high profit margins. When they sell GPU-memory bundles to their AIBs, the profit is added to both the GPU core and the memory, constituting a large portion of the final price of the graphics card. Therefore, even if the bill of materials cost of the memory only increases slightly, manufacturers will pass on this to consumers at a higher rate to maintain their profits.

This is clearly illustrated by comparing the 16GB and 8GB versions of the GeForce RTX 5060 Ti or Radeon RX 9060 XT. The typical bill of materials (BOM) cost for 8GB of video memory is around $20 to $25, but consumers will pay about $50 more for a version with higher memory. This is because graphics card manufacturers take at least a 50% profit margin. With a 50% profit margin, a $25 increase in memory cost translates to a $50 increase in the final selling price.

Of course, this is a simplified explanation. Downstream, there are additional profits, including small profits for partner motherboard manufacturers and profits for distributors and retailers. These profits create a ripple effect, causing a small increase in the BOM cost to translate into a significant increase in the price paid by consumers after passing through the supply chain.

Generally, graphics card prices include a certain buffer to accommodate normal fluctuations in component prices. Earlier this year, a partner motherboard manufacturer told us that the price of GPU + memory bundles fluctuates throughout the year—primarily due to changes in DRAM prices—but most fluctuations are within a manageable range, still reaching the expected manufacturer's suggested retail price.

Occasionally, when prices fluctuate excessively, the supply or price of certain models may be adjusted, but there is usually some flexibility.

However, what the DRAM market is currently experiencing is not "typical price volatility," but rather a significant price surge. According to data from DRAM Exchange, at least so far, GDDR's price increase has not reached the extreme levels of DDR5, but GDDR6 spot prices are rising.

For most of 2025, the average price of GDDR6 was around $2.50 per GB, but the latest report shows that the current price has reached $3.30, an increase of 30%. This trend shows no signs of slowing down, meaning that even greater pressure may be faced in the future.

How rising video memory costs will drive up graphics card prices

A 30% price increase for GDDR memory isn't catastrophic in itself. Our estimates suggest this will raise the bill of materials cost of 16GB GDDR6 from approximately $40 to slightly over $50, while the price of GDDR7, though still uncertain, is likely to cost around $60.

Including profit margins, and assuming neither AMD nor Nvidia reduces profits to maintain price stability, a $10 to $15 increase in memory costs could lead to a $25 to $40 increase in graphics card prices.

These forecasts are based on our modeling, which assumes a 50% profit margin for Radeon products and 60% for GeForce products, while also including various other manufacturing and distribution margins.

Graphics cards with 12GB of memory will be slightly less affected, but a 30% increase in GDDR7 prices could still result in a price increase of at least $25 for products like the RTX 5070. For 8GB memory cards, the impact will be roughly half that of the 16GB models.

The situation would become even more worrying if GDDR prices rose by 50% above normal levels. Most current-generation graphics cards with at least 12GB of VRAM would see their bill of materials (BOM) cost increase by about $20 to $25, which would likely translate into a $50 (or higher) price increase for consumers. To maintain profits, the RTX 5070 could rise to around $600, the RTX 5060 Ti close to $500, and the RX 9060 XT around $400. Even cards with 8GB of VRAM, despite their smaller capacity, could see prices climb to over $300.

The situation would be even worse if VRAM prices doubled, as they have with DDR5 (DDR5 module prices have already doubled). In this scenario, most graphics cards would likely see a price increase of at least $100, as the BOM cost for a typical VRAM configuration would increase by $40 to $50.

This will have the most severe impact on mid-range and low-end models, including the RTX 5060 Ti and RX 9060 XT. Since AMD uses GDDR6 instead of GDDR7, its position might be slightly better, but a $100 price increase would still raise the price of the 9060 XT by about 30%.

With a 100% price increase, the RX 9060 XT could be priced close to the $550 of the RTX 5060 Ti, a similar percentage increase - but such an increase would severely damage the mainstream market. The price increase for high-end graphics cards like the RTX 5080 would be smaller; a $120 increase due to memory costs would only raise the price of that card by about 12%.

And the situation could worsen further. If memory prices triple, graphics card prices could increase by another $200 to $300 - again, assuming profits remain unchanged. By then, the bill of materials (BOM) cost itself will have increased significantly. Even if AMD and Nvidia don't add their usual profit margins and simply pass on the increased memory costs to consumers, graphics card prices will still rise by at least $100. When core component costs surge so dramatically, manufacturers have limited buffer capacity.

Currently, the graphics card market hasn't seen this level of impact, primarily due to time lags. First, graphics card manufacturers sign agreements with memory manufacturers to supply DRAM for specific periods, usually linked to TSMC's GPU core orders. These agreements can cover several months of production needs, meaning that many graphics cards currently on the market likely used memory purchased earlier when prices were lower. Furthermore, there's a natural time lag between component procurement, product assembly, and final retail.

Second, current indicators show that GDDR price increases haven't reached the levels of other types of DRAM - at least not yet. Even if graphics cards are manufactured at current spot prices, the immediate impact on retail prices will be relatively mild.

However, this trend is worrying, with rumors circulating that Nvidia is raising prices with its motherboard partners to offset rising memory costs. There are also reports that models like the RTX 5060 Ti 16GB may face supply constraints.

We cannot confirm these rumors, but as our modeling shows, a significant increase in memory prices will severely weaken the price-performance ratio of mainstream and mid-range 16GB graphics cards. A doubling of memory costs could increase the price of the 5060 Ti 16GB by approximately 30%.

The impact of the release of Nvidia RTX 50 Super series

Recent rumors suggest that the RTX 50 Super series has been delayed or even canceled. While we lack direct information on the status of these models, we can analyze their potential configurations and feasibility in the current DRAM price environment.

For example, regarding the rumored RTX 5070 Super, earlier this year it was reported that the card would increase its memory capacity from 12GB to 18GB by using a 3GB GDDR7 module. With GDDR7 costing approximately $3 per GB, increasing the memory from 12GB to 18GB would raise the bill of materials cost from approximately $36 to approximately $54.

Including Nvidia's profits, this could reasonably increase the card price by about $50. The 5070 Super might launch at $600, while the 5070 at $550; alternatively, Nvidia might price it at $550 while simultaneously lowering the price of the RTX 5070 or discontinuing production.

Even with a mere 30% increase in GDDR7 prices, the bill of materials cost for the 18GB RTX 5070 Super's memory will be nearly double that of the 12GB RTX 5070 at its normal price. To maintain a 60% profit margin, Nvidia might need to raise the manufacturer's suggested retail price by nearly $100, with the actual selling price potentially at least $70 higher.

If memory prices double, Nvidia will face a situation where the actual selling price of the RTX 5070 Super could be about $100 higher than the RTX 5070—potentially making the product uncompetitive. In this scenario, the RTX 5070's price could approach $650, while the 5070 Super with its larger memory would approach $750. This would be a disastrous pricing strategy, especially when consumers compare it to the initial manufacturer's suggested retail price of $550 for the RTX 5070.

It's easy to understand why Nvidia might hesitate to launch the GeForce 50 Super series with larger memory capacity given the fluctuations in the DRAM market. If the main advantage of this series is performance improvement, we might still see these graphics cards released, but if the main selling point is larger memory, then it's not surprising that these plans have been shelved.

Should I buy a graphics card now?

If you're considering buying a new graphics card soon, what does all this mean? Predicting the future is always uncertain, so please treat the following analysis with caution. However, we believe it's unlikely that graphics card prices will fall significantly below manufacturers' suggested retail prices in the coming months.

There may be occasional short-term discounts, but due to rising DRAM prices, newly manufactured models will have higher production costs. It's unclear how long it will take for these cost increases to be reflected in retail prices, but the outlook is not optimistic.

Based on this, if you want the best deal, we recommend buying sooner rather than later - especially if your desired graphics card model is currently priced reasonably.

At the time of writing, the vast majority of graphics cards are selling at manufacturers' suggested retail prices, making this the most favorable time for consumers to buy this year.

Will AMD, Nvidia, or even Intel sacrifice profits to maintain price stability, or at least mitigate the impact of rising DRAM costs? It's possible, but historically, this is unlikely.

Publicly traded companies are very protective of profits because shareholders and executives consider them key performance indicators. Nvidia, with its dominant market position, has little incentive to reduce profits in the gaming sector. AMD could have gained market share by lowering profit margins, but they've always had their own way of doing things.

We've contacted major graphics card manufacturers to inquire about the potential impact of DRAM prices on consumer-grade graphics cards, but unsurprisingly, none have commented yet.

Source: Content from techspot