Video summary
NVIDIA Has GPUs If You Have Coin
Main summary
Key takeaways
Summary of the video’s main points
-
GPU prices are far above MSRP and have worsened over the last 6 months. The channel tracks GPU prices and reports that—excluding one outlier (the 8GB RX 960 XT)—average in-stock prices for current-gen GPUs increased across the board since the previous update.
-
The RTX 5090 is highlighted as the biggest pricing outlier.
- Reported in-stock average for the RTX 5090: ~$4,880, up about $900 in six months.
- This is described as ~144% above MSRP (nearly $3,000 more than MSRP).
- The video argues this is especially severe due to pricing dynamics tied to Nvidia’s memory/VRAM situation and weaker high-end competition.
-
Percent-over-MSRP is rising fastest for GPUs with more VRAM (16GB+).
- For 16GB+ GPUs, the average percent over MSRP allegedly rose to ~67% from ~44% six months earlier.
- For 12GB-or-less GPUs, the average allegedly rose to ~34% from ~23%.
-
Nvidia is blamed as the worst offender.
- The video claims Nvidia GPUs are averaging ~70% over MSRP currently, up from ~39% earlier in the year.
- It contrasts this with AMD (~28%) and Intel (~21%) over MSRP.
-
Root cause: VRAM/memory pricing and supply constraints—plus Nvidia’s tighter control of the bundle.
- The video explains that while Nvidia/AMD don’t manufacture memory chips, they package and sell the GPU + VRAM as a unit, giving them leverage when memory prices rise.
- It cites reporting that Nvidia has raised prices of GDDR6 and GDDR7 passed through to board partners (AIBs), contributing to another round of graphics-card cost increases.
- It also references a quote attributed to industry reporting about rising costs per gigabyte for GDDR7 modules.
-
High-end product timing is portrayed as being influenced by memory costs.
- The video discusses rumors/reporting about delays to “Super” refresh cards (RTX 50-series “Super”), claiming at least one board partner may already have received some units, but release is delayed again due to VRAM pricing.
- It also references commentary suggesting 2026 may be the first year in decades without a new Nvidia consumer GPU generation, at least regarding launch cadence (with caveats about how “generation” is defined).
-
Competitor vs. market behavior: AMD is said not to collude with Nvidia, but prices still rise.
- The video emphasizes that Nvidia and AMD compete rather than coordinate, framing the price increases as structural rather than cartel-like.
-
AIB partners and pricing opacity are discussed, but the core argument stays focused on memory and vendor control.
- The video cites warnings that PC demand is slowing due to component cost increases, and that availability will likely worsen in the second half of 2026.
- It references research suggesting AIB selling prices rose faster than memory-cost deltas, implying partners may capitalize on scarcity/press attention.
- Even so, it repeatedly returns to VRAM cost as the primary driver.
-
Methodology and market snapshot claims.
- Prices are based on in-stock listings from Newegg for first-party listings across current GPUs.
- Out-of-stock listings are excluded, with the argument that in-stock data better reflects what consumers can actually buy.
-
Longer-term view: prices may stay bad or worsen, with only an AI bubble bust potentially improving things.
- The video argues the market has structurally changed and that next-gen GPU launch prices may be higher.
- The only scenario offered for meaningful VRAM improvement is an AI-market bubble/pop, but it warns that would likely bring broad economic harm rather than a “good” consumer reset.
-
Practical consumer advice: avoid building gaming PCs unless you must.
- The channel advises against building a gaming PC now except in necessity cases.
- Suggested alternatives include using older GPUs, harvesting parts, and waiting for better conditions.
Presenters / contributors
- Gamers Nexus (video host/speaker; referenced throughout as “we” and via the Gamers Nexus brand)
- Morgan Freeman (mentioned as a hypothetical narrator style; not treated as an actual contributor)
- Jensen (referenced for a quote, not as an on-screen contributor)