The video is the first part of Hardware Unboxed's November Q&A, with Tim and Steve answering viewer questions. It opens with a sponsor segment for Thermal Grizzly's WireView Pro II, a device that sits between graphics card and power cable to monitor current, voltage and temperature per pin.

The first question asks whether a 24 GB RTX 5080 will get an MSRP. The hosts point to the RTX 3080 12 GB, which they say arrived during the crypto boom without a stated price. They expect little change beyond memory, since the GB203 die is already fully enabled, and say the card would be a niche, expensive product.

The title question, whether AMD missed an opportunity with a faster RDNA 4 card, gets the most considered answer. Both hosts stress it is speculation. Their view is that AMD read the weak RDNA 3 reception as a reason to avoid the top tier, judged that a card between the RTX 5080 and 5090 would struggle on price, saw a near 600 W rival coming, and preferred to spend wafers on smaller dies. They add that FSR 4 and other features might have changed that call in hindsight.

Next they criticise Sam Altman's remarks about memory being better used in data centres than in home PCs, calling the comments tone deaf while conceding the utilisation point. A long discussion of whether a GPU faster than the RTX 5090 will arrive follows. They see a 6090 as likely to be planned but uncertain to reach gamers at a sensible price.

The last question covers CPU bottlenecks. The hosts argue higher refresh rates, upscaling and ray tracing push more load onto the CPU, and that a Zen 6-based PS6 could raise the baseline developers target.

Key takeaways

  • The hosts say AMD's decision to skip a higher-end RDNA 4 card was probably commercial, shaped by RDNA 3 sales, a near 600 W rival and wafer use, though they stress it is speculation.
  • They expect a 24 GB RTX 5080 to be the same GB203 silicon with more memory, so no extra cores or bandwidth, and possibly launched with no formal MSRP.
  • On Sam Altman's comments, they accept the utilisation argument in narrow terms but call the remarks self-serving and badly timed.
  • They are unsure a consumer GPU faster than the RTX 5090 will appear soon, because AI demand may take the silicon or price it beyond gaming.
  • They argue chasing higher frame rates, upscaling and ray tracing make the CPU matter more, and a Zen 6-based PS6 could lift the baseline.

Our take

The AMD discussion is informed opinion, and the hosts say as much: Tim says he has not asked AMD about it. The reasoning on pricing, power and wafer use is plausible but unconfirmed, so treat it as analysis from a single outlet. Pricing and availability remarks reflect the state of the market when the video was made.

The CPU section is the most practically useful for buyers, since it explains why a slow older chip can hold back a fast GPU at high frame rates. None of our shortlisted coverage bears directly on the GPU questions here, so we draw no comparison.