News
IndustryTSMC’s grip on advanced chipmaking will face competitionIndustryCadence joins European Chip Design Platform to support semiconductor startups and SMEsMaking chipsMarc Hijink on ASML’s Culture, Supply Chain, and GeopoliticsProcessorsIntel's post-Nova Lake CPUs take shape as Razor Lake desktop and laptop variants surfaceAI chipsHow Google’s New AI Chip Software Helps It Battle NvidiaIndiaMore IC Packaging Plants Emerge In IndiaIndustryUS Tightens China AI Chip Access as Nvidia, TSMC and Semiconductor Stocks SlideDesignAxelera AI moves into HPC with TitaniaMaking chipsChinese envoy says chip equipment export curbs threaten EU-China trade thawIndiaSEMICON India 2026 Day 2 Sees 25 MoUs, Strategic Collaborations Across Semiconductor Value ChainIndustryTSMC’s grip on advanced chipmaking will face competitionIndustryCadence joins European Chip Design Platform to support semiconductor startups and SMEsMaking chipsMarc Hijink on ASML’s Culture, Supply Chain, and GeopoliticsProcessorsIntel's post-Nova Lake CPUs take shape as Razor Lake desktop and laptop variants surfaceAI chipsHow Google’s New AI Chip Software Helps It Battle NvidiaIndiaMore IC Packaging Plants Emerge In IndiaIndustryUS Tightens China AI Chip Access as Nvidia, TSMC and Semiconductor Stocks SlideDesignAxelera AI moves into HPC with TitaniaMaking chipsChinese envoy says chip equipment export curbs threaten EU-China trade thawIndiaSEMICON India 2026 Day 2 Sees 25 MoUs, Strategic Collaborations Across Semiconductor Value Chain
EveryChip
Video from Geekerwan about the technologies of the new NVIDIA GeForce RTX 50 generation. Here you can see the graphics cards.

GeForce RTX 50 series

Nvidia's Blackwell consumer graphics cards, from the RTX 5050 to the RTX 5090

Photo: 极客湾Geekerwan, CC BY 3.0, via Wikimedia Commons

The GeForce RTX 50 series is Nvidia's consumer graphics card line built on the Blackwell microarchitecture, announced at CES 2025 and first shipped in January 2025. Made by TSMC on a custom 4N process, it introduced GDDR7 memory, fourth-generation RT cores, fifth-generation Tensor Cores and DLSS 4 to Nvidia's desktop and laptop GPUs.

Share on WhatsApp

  • Announced at CES 2025 on January 6, with the first cards released that month
  • First consumer GPUs to use GDDR7 video memory
  • Built by TSMC on a custom 4N 5 nm process
  • DisplayPort 2.1b UHBR20 output at 80 Gbps
  • DLSS 4 Multi Frame Generation is exclusive to this series

What it is

The GeForce RTX 50 series is Nvidia's line of consumer graphics cards, covering desktop and laptop GPUs from the entry-level RTX 5050 up to the enthusiast RTX 5090. Announced at CES 2025 in Las Vegas on January 6, 2025, the first cards went on sale later that month. The chips use Nvidia's Blackwell microarchitecture and carry fourth-generation RT cores for hardware-accelerated real-time ray tracing plus fifth-generation Tensor Cores aimed at AI work. TSMC builds the GPUs on a custom 4N process. The series is the first consumer line to use GDDR7 video memory and the second to offer a PCIe 5.0 interface.

How it came about

Nvidia revealed Blackwell for datacentre products in March 2024. Unlike the split between Ada Lovelace and Hopper, one architecture now serves both consumer and server parts. A cinematic trailer for The Witcher IV shown at the Game Awards in December 2024 was rendered on hardware Nvidia had not yet named; after the launch the company confirmed an RTX 5090 had done the work in Unreal Engine 5. Reports that month said units were being stockpiled in US warehouses ahead of threatened import tariffs. The desktop and mobile line was formally announced at the CES keynote on January 6, 2025.

Inside the chip

Blackwell keeps Ada Lovelace's approach of high graphics clock speeds paired with a large L2 cache, the small on-chip memory that sits close to the cores. RT cores handle the maths behind ray tracing, which traces light paths for realistic shadows and reflections, while Tensor Cores accelerate the matrix operations used by deep learning. Desktop cards use GDDR7 memory, giving more bandwidth than the GDDR6 and GDDR6X of the previous generation over the same bus width; Samsung supplied the modules because they cleared validation first. Every design must use the 16-pin 12V-2x6 power connector, a revised 12VHPWR part whose shortened sense pins stop power flowing until it is fully seated. Outputs include DisplayPort 2.1b UHBR20 at 80 Gbps, and the media engine adds ninth-generation NVENC encoding and sixth-generation NVDEC decoding with 4:2:2 colour support.

How it was made

TSMC manufactures the GPUs on a custom 5 nm process that Nvidia calls 4N. The series is the second consumer family to use a PCIe 5.0 interface and the first to carry GDDR7 memory, apart from the entry RTX 5050, which stays on GDDR6. Transistor counts and package details are not given here. Nvidia designs the chips and markets the cards, with add-in board partners producing their own versions.

GeForce RTX 50 series
A drawing made by EveryChip from what this page says. It is a diagram, not a picture of the real chip.

Where it went

The cards target desktop PCs and gaming laptops. Laptop models were shown at CES 2025 and paired with Intel Arrow Lake-HX or AMD Strix Point and Fire Range processors; the first machines arrived in March 2025 from $1,299. Max-Q power features such as Advanced Power Gating, which switches off unused areas of the GPU, are claimed to extend battery life by up to 40% over GeForce 40 series laptops. The RTX 5090D variant is sold only in the Chinese market. DLSS 4 Multi Frame Generation launched with 75 titles, among them Alan Wake 2, Cyberpunk 2077 and Star Wars Outlaws.

Versions and family

The line runs from the entry RTX 5050 and RTX 5060, through mid-range RTX 5060 Ti cards with 8 GB or 16 GB, to the high-end RTX 5070, 5070 Ti and 5080, and the enthusiast RTX 5090. A 5090D is made for China. The series replaces the GeForce RTX 40 series. Laptop versions of the same GPUs were announced alongside the desktop parts.

Why it matters

The series brought GDDR7 and DisplayPort 2.1b to mainstream graphics cards and tied consumer hardware to the same Blackwell architecture used in Nvidia's datacentre products. It also introduced DLSS 4, whose Multi Frame Generation can produce several interpolated frames from one rendered frame, a capability reserved for these cards. Later DLSS updates kept the line central: DLSS 4.5 added a 6x generation mode exclusive to the series, and DLSS 5, which arrived in September 2026, runs only on RTX 50 hardware.

Problems and reception

Launch supply was tight and prices ran well above MSRP, with a memory shortage worsening matters late in 2025. The 12V-2x6 connector melted on some 5090 and 5080 cards, and some dies shipped missing eight ROPs, units that write pixels to memory, which Nvidia said affected under 0.5% of cards. Black screen boots, unstable Windows driver branches and the loss of 32-bit CUDA, OpenCL and PhysX support drew complaints, though PhysX returned for some games in December 2025. Reviewers also disputed performance claims built on Multi Frame Generation, and the RTX 5090 drew a lukewarm reception.

Questions people ask about GeForce RTX 50 series

What is the GeForce RTX 50 series?

It is Nvidia's consumer graphics card family based on the Blackwell microarchitecture, announced at CES 2025 and released from January 2025. It replaces the GeForce RTX 40 series and covers desktop and laptop GPUs from the RTX 5050 up to the RTX 5090.

Who makes the chips?

Nvidia designs the GPUs and TSMC manufactures them on a custom 4N process node, a 5 nm class process. Nvidia also markets the Founders Edition cards, while add-in board partners build their own versions.

What does GDDR7 change?

GDDR7 is the video memory used on these cards, giving more bandwidth than the GDDR6 and GDDR6X of the GeForce 40 series over the same bus width. Desktop cards use Samsung modules, and the entry RTX 5050 still uses GDDR6.

What is Multi Frame Generation?

It is part of DLSS 4 and generates several extra frames from each traditionally rendered one, whereas the GeForce 40 series could add only one. It is exclusive to the RTX 50 series, though reviewers note it raises input latency and can show artefacts at low base frame rates.

What problems have been reported?

Owners have seen melting 12V-2x6 connectors, black screen boots after driver installs, and some cards shipped with eight missing ROPs. Supply was scarce at launch and prices stayed above MSRP, and 32-bit CUDA and PhysX support was dropped before returning for some games.

Good reading and sources

Where this page comes from

The facts on this page come from the Wikipedia article "GeForce RTX 50 series", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryChip rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.

It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the chip of the day, new pages on EveryChip, and the conferences and anniversaries of the week ahead.