The September 26, 2024 leak was close on the RTX 5090’s core count and memory configuration, but not on its final power rating. NVIDIA’s shipping GeForce RTX 5090 has a 575W total graphics power rating—not 600W—with 21,760 CUDA cores, 32GB of GDDR7 memory, and a 512-bit memory interface. The RTX 5080 also came in below the rumor: 360W rather than 400W.
The September 26, 2024 leak was close on the RTX 5090’s core count and memory configuration, but not on its final power rating. NVIDIA’s shipping GeForce RTX 5090 has a 575W total graphics power (TGP) rating—not 600W—with 21,760 CUDA cores, 32GB of GDDR7 memory, and a 512-bit memory interface. The RTX 5080 also came in below the rumor: 360W rather than 400W, alongside 10,752 CUDA cores, 16GB of GDDR7, and a 256-bit interface.
In other words, the leak correctly anticipated several important specifications, but its headline wattage figures were pre-launch estimates rather than final product specifications.
As an Amazon Associate I earn from qualifying purchases.
Recommended Free Tools
RTX 5090 versus RTX 5080: leaked and confirmed specifications
| Specification | GeForce RTX 5090 | GeForce RTX 5080 |
|---|---|---|
| Architecture | Blackwell | Blackwell |
| CUDA cores | 21,760 | 10,752 |
| Memory | 32GB GDDR7 | 16GB GDDR7 |
| Memory interface | 512-bit | 256-bit |
| Official total graphics power | 575W | 360W |
| NVIDIA-listed system power | 850W minimum | 850W minimum |
The confirmed specifications come from NVIDIA’s product pages and Blackwell architecture documentation. The 5090’s listed boost clock is 2.41GHz, while the 5080 is listed at 2.62GHz. Those clock figures should not be used by themselves to predict gaming performance: architecture, cooling, power limits, memory bandwidth, software features, and the workload all matter.
What the original RTX 5090 leak claimed
Tom’s Hardware reported on September 26, 2024, that a pre-release rumor attributed a 600W TGP to the RTX 5090 and a 400W TGP to the RTX 5080. The same rumor package described a 5090 with as many as 21,760 cores, 32GB of memory, and a 512-bit bus.
#1 Best Overall
- Powered by the NVIDIA Blackwell architecture and DLSS 4. System Requirements: Minimum 850W PSU with 16-pin 12V-2x6 (12VHPWR) connector required. Verify before purchasing.
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability. Compatibility: 348mm (13.7") length, 3.6 slots, 4.3 lbs. Confirm case clearance and slot spacing. GPU bracket included.
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.6-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
That rumor was unusually accurate on the RTX 5090’s major architectural and memory details:
- 21,760 CUDA cores: confirmed.
- 32GB of memory: confirmed as GDDR7.
- 512-bit memory interface: confirmed.
- 600W TGP: not the final official rating; NVIDIA lists 575W.
NVIDIA’s Blackwell documentation identifies the RTX 5090 as a GB202-based product with 170 streaming multiprocessors, 680 fifth-generation Tensor Cores, and 170 fourth-generation ray-tracing cores. NVIDIA also announced 1,792GB/sec of memory bandwidth for the card.
For readers comparing the shipping product, the GeForce RTX 5090 is the relevant specification—not the earlier 600W rumor. Exact board dimensions, cooler design, clock speeds, connector arrangements, and power recommendations can differ between NVIDIA’s Founders Edition and partner add-in-board models.
What the leak got right and wrong about the RTX 5080
The rumor also anticipated much of the RTX 5080’s eventual configuration. NVIDIA’s final product page lists:
Rank #2
- Powered by the NVIDIA Blackwell architecture and DLSS 4 OC mode: 2640MHz/Default mode: 2610MHz (Boost Clock)
- Military-grade components deliver rock-solid power and longer lifespan for ultimate durability
- Protective PCB coating helps protect against short circuits caused by moisture, dust, or debris
- 3.125-slot design with massive fin array optimized for airflow from three Axial-tech fans
- Phase-change GPU thermal pad helps ensure optimal thermal performance and longevity, outlasting traditional thermal paste for graphics cards under heavy loads
- 10,752 CUDA cores, matching the later reported specification.
- 16GB of GDDR7.
- A 256-bit memory interface.
- 360W TGP, rather than the leaked 400W figure.
The RTX 5080 therefore has half the RTX 5090’s listed memory capacity and half its memory-bus width. It also has 84 Blackwell streaming multiprocessors compared with 170 in the RTX 5090. That is a substantial product-tier difference, not simply a small clock-speed variation between two otherwise similar cards.
NVIDIA listed a $999 starting price and January 30 availability for the RTX 5080 at launch. Those were launch-reference figures, not guarantees of current retail pricing, stock, or partner-card pricing.
Why 575W versus 600W still matters
TGP is a board-level power target or limit used for graphics-card and system planning. It does not mean the GPU draws exactly that amount continuously in every game or application. Actual consumption varies with workload, frame-rate limits, voltage, clocks, temperature, firmware, and the specific card model.
Even so, a 575W flagship is a serious system-design consideration. The confirmed power figures affect:
Rank #3
- Powered by the NVIDIA Blackwell architecture and DLSS 4
- Powered by GeForce RTX 5080
- Integrated with 16GB GDDR7 256bit memory interface
- PCIe 5.0
- WINDFORCE cooling system
- Power-supply capacity: NVIDIA lists an 850W minimum system-power recommendation for both the Founders Edition RTX 5090 and RTX 5080 product pages.
- Power cabling: the RTX 5090 quick-start documentation specifies a 16-pin PCIe Gen 5 power connection and recommends using the supplied NVIDIA adapter with the Founders Edition.
- Case clearance: NVIDIA’s RTX 5090 guide documents a Founders Edition size of 304mm long, 137mm wide, and 61mm high.
- Thermal planning: a high-power card needs appropriate case airflow and enough physical room around its cooler.
Do not assume that any generic adapter is interchangeable. Follow the exact PSU, cable, graphics-card, and add-in-board manufacturer instructions. The connector and cable requirements for a partner RTX 5090 may not be identical to those for the Founders Edition.
Does the RTX 5090 simply double the RTX 5080?
On the headline specifications, the gap is unusually large. The RTX 5090 has 170 streaming multiprocessors and 21,760 CUDA cores, compared with 84 streaming multiprocessors and 10,752 CUDA cores in the RTX 5080. It also doubles the listed memory capacity and memory-bus width.
Free tools Windows power users keep installed
One-click scans. No signup required.
That does not establish a universal two-times gaming performance advantage. The practical difference depends on resolution, ray-tracing settings, the game engine, CPU limitations, DLSS settings, frame generation, and whether the workload is constrained by compute, memory capacity, or bandwidth. Benchmark results should be evaluated by game and resolution rather than inferred directly from the core-count ratio.
The 5090’s 32GB of GDDR7 may be particularly relevant to creators and people running local AI workloads, where fitting a larger project or model into GPU memory can matter more than a simple shader-count comparison. This specification alone does not prove a particular application-level speedup, however.
Rank #4
- Chipset: GeForce RTX 3050
- Boost Clock / Memory: 1492 MHz / 14 Gbps
- Video Memory: 6GB GDDR6
- Memory Interface: 96-bit
- Output: DisplayPort x 1 (v1.4a) / HDMI 2.1a x 2
What buyers should check before upgrading
- Confirm the exact model. Specifications vary by add-in-board manufacturer. Check the product page for the specific RTX 5090 or RTX 5080 rather than relying on a generic series specification.
- Measure the case. Check GPU length, height, slot thickness, radiator placement, front fans, and cable-bend clearance. The 304mm by 137mm by 61mm dimensions above apply to the referenced RTX 5090 Founders Edition, not automatically to every partner card.
- Verify PSU capacity and cabling. Start with NVIDIA’s listed 850W system-power recommendation, then account for the rest of the system and the partner card’s requirements. Confirm that the PSU and cable arrangement comply with the card maker’s instructions.
- Plan airflow. A high TGP increases the importance of intake, exhaust, fan placement, and unobstructed cooler clearance. A card that physically fits may still be a poor fit for a cramped or poorly ventilated case.
- Match the card to the workload. High-resolution ray-traced gaming, demanding creator applications, and memory-heavy local AI tasks can justify the 5090’s larger memory configuration. For less demanding gaming workloads, the 5080’s lower power target and lower launch price may be more relevant.
The leak’s final accuracy verdict
The best summary is that the RTX 5090 leak was highly accurate about the product’s core and memory architecture but overstated its final power rating. The RTX 5090 shipped with 21,760 CUDA cores, 32GB of GDDR7, and a 512-bit bus, while its official TGP settled at 575W.
The RTX 5080 likewise shipped with the rumored 10,752 CUDA cores, 16GB of GDDR7, and 256-bit interface, but its official TGP was 360W rather than 400W. For upgrade planning, the official NVIDIA specifications—not the original rumor—should control decisions about power supplies, connectors, dimensions, and case compatibility.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Frequently Asked Questions
Does the RTX 5090 really have a 600W TGP?
No. NVIDIA’s official RTX 5090 product specifications list a 575W total graphics power rating. The 600W figure came from a September 2024 pre-launch leak.
Is the RTX 5080 rated at 400W?
No. The RTX 5080’s official TGP is 360W. The 400W figure was part of the same pre-launch rumor package.
What power supply does an RTX 5090 or RTX 5080 need?
NVIDIA lists an 850W minimum system-power recommendation for both the RTX 5090 and RTX 5080 Founders Edition product pages. The exact PSU and cable requirements should be checked against the specific card and power supply.
The Bottom Line
Bottom line: the leak got the RTX 5090’s 21,760 cores, 32GB of GDDR7, and 512-bit bus right, but the final official rating is 575W instead of 600W. The RTX 5080 also landed below its rumored figure at 360W instead of 400W, with 10,752 cores, 16GB of GDDR7, and a 256-bit interface.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
