Expensive GDDR7 VRAM expected to push RTX 5090 and 5080 graphics cards above MSRP

Daniel Sims

Posts: 1,876   +49
Staff
TL;DR: A graphics card's MSRP often serves more as a guideline than a fixed rule, as hardware partners frequently release higher-priced custom variants. With concerns growing over limited supplies of Nvidia's upcoming RTX 5090 and 5080, reports suggest that obtaining these cards at their manufacturer's suggested retail price may be even more challenging than in previous generations.

A leaker on the Chiphell forums recently suggested that Nvidia's board partners will struggle to sell custom RTX 50 series graphics cards at MSRP due to the high cost of GDDR7 memory. Profit margins are expected to be lower than usual.

As consumers eagerly anticipate the launch of the RTX 5090 and 5080 later this week, the Nvidia-manufactured Founders Edition variants have garnered the most attention. Although the company sells its in-house GPUs at standard retail pricing, they are expected to sell out quickly, leaving partner cards with various cooling and overclocking configurations as the only option.

MSI, Gigabyte, and other vendors have revealed numerous custom options, but most GPU partners have yet to disclose pricing information. Almost all of the few price tags revealed thus far are hundreds of dollars above MSRP.

Several Gigabyte RTX 5080 models are currently listed on Best Buy, but only one, the Windforce, matches the $999 Founders Edition. Other options range between $1,199 and $1,399. Meanwhile, leaked prices from B&H indicate that Asus, MSI, and Gigabyte will sell the GPU for between $1,299 and $1,699. The flagship RTX 5090 starts at $2,000 but partner prices sit between $2,379 and $2,799.

A tipster on the Chiphell forums indicated that GDDR7 VRAM, included in every RTX 50 series GPU, is the main culprit. The new memory modules are faster but costlier than the GDDR6 VRAM used in RTX 40 series, Intel Arc Battlemage, and AMD Radeon RX 9000 cards, and the leaker suspects that board partners are shouldering the expense. In this scenario, the leaker said that selling RTX 50 GPUs is "like charity."

The RTX 5090 and 5080 are set to launch on January 30, and consumers have already begun camping outside Micro Center days beforehand. The RTX 5070 Ti and 5070 are expected to arrive in February for $749 and $549 respectively. Details on the RTX 5060 are forthcoming.

Additionally, AMD's Radeon RX 9070 XT and 9070 begin shipping in March, but their prices remain uncertain. Team Red's next-generation GPUs are expected to compete with the RTX 5070.

Permalink to story:

 
I mean, look above: I think Nvidia would want to offset the increase to get in some sales after they basically got done in AI yesterday.
 
Another day and the wind blows south and GPU prices go North. Nothing new, moving on...
 
Huh, looks like RDNA 4 has a chance after all if they can stick closer to MSRP. Real prices in the wild might be many hundred dollars apart and AIBPs will be happy.
 
The 4080 didn't sell well at $1200 so I don't know how they think the 5080s priced $1400-1900 will sell.

I'll be interested to see what prices are like in a few months when the hype has died down. But right now it's looking like the 5070 TI is going to be the only sane choice between VRAM and pricing.
 
The 4080 didn't sell well at $1200 so I don't know how they think the 5080s priced $1400-1900 will sell.

I'll be interested to see what prices are like in a few months when the hype has died down. But right now it's looking like the 5070 TI is going to be the only sane choice between VRAM and pricing.
Well, that's mostly nVidia charging board partners so much that it isn't profitable unless they charge massive amounts over MSRP.

The MSRP is and always will be a lie. AMD will be able to sell the 9070 XT at $600 because the real world price of the 5070 is going to be ~$800.
 
When you look at the technology in the 5090 -32GB GDDR7 giving it almost twice the bandwidth, new and more Tensor Cores and RT Cores, vapor chambers, liquid metal etc etc it's surprising how little of a performance uplift there is over the 4090...
 
This is how nVidia becomes the sole AIB while saving face about GPU costs. They can just blame their partners' higher prices on memory costs. EVGA is probably glad they got out when they did.

- I figure we're well on our way to seeing Nvidia partner with purchase a single AIB (likely PNY) and exclusively use them to bring cards to market.

As far as NV is concerned the AIBs are more like parasites than a value add, why let an AIB pocket the extra change when NV could pocket it instead?
 
If only we could convince the ******* whales to leave them on the shelves, gathering dust...Unfortunately, they (and no doubt scalpers) are already camping out in front of stores as we speak....oh well........."There's one born every minute"..........sad, sad,sad,,,,,,,,,,,,,
 
Back