RTX 5080 and RTX 5070 Prices in November 2026: Buy Now or Wait?

The RTX 5080 costs about $2,000 in November 2026, double its $999 launch MSRP, while the RTX 5070 and RTX 5060 Ti 16GB sit about 64% and 79% above theirs. The cheapest US listings in Tom's Hardware's GPU tracker, last updated November 3, were $1,999 for the RTX 5080 and $899 for the RTX 5070. Prices have climbed through the fall, and we found no announcement of a new Nvidia MSRP. This guide gives the reported prices with their dates, explains which card suits AI and CAD work by VRAM, and sets out when buying now makes sense. For the flagship card, see our separate guide to RTX 5090 prices. All prices below are reported street figures, not Nexus Compute prices.

What are RTX 5080, 5070 and 5060 Ti prices in November 2026?

These are the lowest US prices in Tom's Hardware's tracker as of November 3, 2026, against launch MSRP. The multiples are our arithmetic.

• RTX 5080 16GB: $1,999 against $999 MSRP (2.0x).
• RTX 5070 Ti 16GB: $1,399 against $749 MSRP (1.9x).
• RTX 5070 12GB: $899 against $549 MSRP (1.6x).
• RTX 5060 Ti 16GB: $769 against $429 MSRP (1.8x).
• RTX 5060 Ti 8GB: $499 against $379 MSRP (1.3x).

Other reports differ, so treat the RTX 5080 as a range of roughly $1,980 to $2,500. A VideoCardz survey on October 28 found $1,979.99 at Micro Center for local pickup and $1,999.99 at Best Buy online (Notebookcheck). Tom's Hardware put the cheapest in-stock card at $2,199.99 on October 30 (Tom's Hardware). Hardware Unboxed's October survey listed a US minimum of $2,200, up from $1,580 in September (MIXED), and Club386 cited prices of about $2,500 in the US on November 3 (Club386). The spread reflects timing, model and stock: the Notebookcheck report found one Asus Prime OC priced $270 above the higher-tier Astral OC at the same store.

Why can you not buy these cards at MSRP?

MSRP has become a reference point rather than a price. Tom's Hardware reported on October 30 that the RTX 5080 had disappeared from Nvidia's marketplace after the RTX 5090, while the RTX 5070 Ti and RTX 5070 remained listed as out of stock. The same report said the removal does not necessarily mean the card is discontinued, and that talk of a replacement RTX 5080 24GB is still a rumor. An earlier sign of the squeeze came in August, when PC Gamer found the RTX 5060 Ti 16GB at $780, above the cheapest RTX 5070 at $763 (PC Gamer). That inversion has since reversed on the Tom's Hardware tracker, which shows how quickly the order of these cards can change.

Why are RTX 50 prices still rising?

Memory is the main reason. Tom's Hardware reported on October 29 that memory makers have moved capacity toward high-margin HBM for AI data centers, which limits the memory available for graphics cards (Tom's Hardware). More increases may follow: Club386 relayed unconfirmed claims from a Chinese industry channel on November 3 that Colorful would raise the RTX 5080 by 2,500 to 2,800 yuan, about $400, and the RTX 5070 Ti by 1,000 yuan. Do not expect an RTX 50 Super refresh to reset the market soon. Club386 reported on July 17, citing VideoCardz, that the Super cards were on hold because 3GB GDDR7 chips cost up to three times as much as 2GB chips (Club386). Nvidia has not announced those cards.

Which RTX card suits AI work by VRAM?

For local AI, VRAM decides what fits, and speed decides how long it takes. As a rule of thumb, a model needs about 1GB per billion parameters at 8-bit and about 0.5GB at 4-bit, plus room for context, so a 30-billion-parameter model at 4-bit needs about 15GB before overhead. Our guides to VRAM sizing for LLMs and the best hardware for local LLMs go further. NVIDIA lists these memory sizes:

• RTX 5080 16GB: 16GB on a 256-bit bus (NVIDIA). The fastest card here, for buyers who run the same 16GB-sized jobs many times a day. Also available as the Zotac Solid OC.
• RTX 5070 Ti 16GB: 16GB on a 256-bit bus. On the November 3 tracker it was $600 below the RTX 5080 with the same capacity, so it is the value pick when 16GB is the target.
• RTX 5070 12GB: 12GB on a 192-bit bus. Fine for smaller models and image generation, but 12GB is the limit that buyers outgrow first.
• RTX 5060 Ti 16GB: 16GB on a 128-bit bus (NVIDIA). The cheapest 16GB GeForce on the tracker at $769. It runs slower, but it loads the same size of model. See also the Ventus 3X.
• RTX 5060 Ti 8GB: avoid it for AI work. A saving of $270 is not worth losing half the memory.

If 16GB will not hold your models, none of these cards is the answer. Look at the RTX PRO range, or at renting, since no GeForce under the RTX 5090 has more than 16GB. AMD's RX 9060 XT 16GB and RX 9070 XT were $659 and $919 on the same tracker, but check that your AI tools support them before saving the difference.

Which RTX card suits CAD work?

Most CAD buyers need less than these prices imply. Our RTX PRO VRAM guide notes that parametric modeling and typical assemblies rarely approach 24GB, so a 12GB or 16GB card may cover many seats. Check your largest assembly first. The real question is support. If your CAD vendor or employer requires a certified professional card, a GeForce may not qualify, and the RTX PRO 4000 Blackwell 24GB is the entry point. Check your vendor's hardware list first, and see our guide to ISV-certified drivers. For rendering and simulation, where GPU memory fills quickly, buy 16GB at least.

Should you buy an RTX 5080 or 5070 now or wait?

• Buy now if the work is funded and a delay costs more than the premium. Tom's Hardware said on October 30 that the RTX 5080 is unlikely to be available by Black Friday at current rates.
• Wait if the purchase is optional and your current GPU still does the job. Nothing we found points to a price drop, but paying 2x MSRP for a card you do not need yet is a poor use of budget.
• Rent if the need is short. Paying by the hour for a few weeks of work can beat buying at 2x MSRP, and our neocloud rental guide covers the trade-offs.
• Step down a tier if your jobs fit. Moving from the RTX 5080 to the RTX 5070 Ti keeps 16GB and saved $600 on the November 3 tracker.
• Do not count on Black Friday. Our Black Friday price guide explains what to expect this year.

What should you watch for when buying at these prices?

Who sells the card matters more now. Tom's Hardware reported on October 29 that much of the online RTX 5080 stock is with third-party sellers, including one Amazon seller with an 81% positive rating. Check who honors the warranty, whether the card is new, and what the return terms are. Quotes for high-demand GPUs can also expire quickly, so get the validity period in writing. Finally, check that the card fits: length, slot width, power connector and power supply all matter in a workstation chassis, and our GPU fit guide covers each one.

Nexus Compute supplies new graphics cards and complete AI workstations through authorized distribution channels with the full manufacturer warranty, and checks compatibility against your platform before dispatch. We do not publish prices or stock in this guide because they change quickly. Request a quote with your workload and chassis and we will reply within 48 business hours, including when waiting or renting looks like the better answer.