Ada Lovelace · Ada Data Center
AcceleratorsGPU0054
NVIDIA
L20 48GB
No tracked market supply yet. Kovara does not currently track public offerings for this accelerator.
Accelerator Overview
Core hardware and lifecycle, drawn from Kovara master data.
Core hardware
- Architecture
- Ada Lovelace
- Generation
- Ada Data Center
- VRAM
- 48 GB
- Memory type
- GDDR6
- Memory bandwidth
- 0.864 TB/s
- Form factor
- PCIe dual-slot FHFL
Lifecycle / usage
- Primary use
- AI inference / graphics / China-market compute
- Launch year
- 2024
- Generation status
- Current
- Last verified
- 03 Sept 2026
The hardware in depth
What the recorded specifications mean in practice, and what this accelerator is used for.
Specs, explained
Used for AI inference, graphics and China-market compute
- VRAM
- 48 GB
- Memory type
- GDDR6
- Memory bandwidth
- 0.864 TB/s
GDDR is graphics memory mounted on the board around the GPU rather than stacked on the package. It is widely used on graphics, rendering and cost-efficient inference cards.
VRAM is the working space on the card. Model weights, activations and datasets have to fit in it, or be split across several GPUs. Rule of thumb for the weights alone at 16-bit precision (about 2 GB per billion parameters): up to roughly 24 billion parameters, before activations and cache. This is arithmetic on the recorded VRAM, not a benchmark.
- Architecture
- Ada Lovelace
- Generation
- Ada Data Center
- Launch year
- 2024
- Lifecycle status
- Current
Ada Lovelace brought fourth-generation Tensor Cores with FP8 support and third-generation ray-tracing cores. (General context about the architecture, not a statement about this specific card.)
Launched in 2024. Kovara's records list it as a current-generation part.
- NVLink
- Not supported
- MIG
- Not supported
- Form factor
- PCIe dual-slot FHFL
- TDP
- 275 W
Kovara's records show no NVLink support, so multi-GPU jobs on this part rely on PCIe or the network between servers to move data.
PCIe cards slot into standard server and workstation expansion slots, which suits a wide range of systems.
TDP is the rated thermal design power: the planning figure for power delivery and cooling in a server.
- Primary use (recorded)
- AI inference / graphics / China-market compute
These are the uses Kovara has recorded for this part. The notes explain the workload in general terms; they are not performance claims.
- AI inferenceRunning a trained model to answer requests. Memory capacity, latency and cost per request are the usual concerns.
- graphicsInteractive graphics and professional design or engineering applications.
- China-market compute
Figures are Kovara master data, last verified 03 Sept 2026. Explanations describe the technology in general; they are not benchmarks and say nothing about price or availability.
Explore this accelerator on Compute- GDDR6 chips
- GPU die
- Substrate
- Board
- Power
- Input
- Output
- PCIe connector
Market Intelligence
Market pricing is not yet tracked for this accelerator.
Tracked coverage does not describe all market supply or geographic availability.
Explore Compute →Technical Specifications
Recorded hardware attributes. Not tracked is distinct from No.
Compute
- Architecture
- Ada Lovelace
- Generation
- Ada Data Center
- Primary use
- AI inference / graphics / China-market compute
Memory
- VRAM
- 48 GB
- Memory type
- GDDR6
- Memory bandwidth
- 0.864 TB/s
Platform
- Form factor
- PCIe dual-slot FHFL
- TDP
- 275 W
- NVLink supported
- No
- MIG supported
- No
Lifecycle / usage
- Primary use
- AI inference / graphics / China-market compute
- Launch year
- 2024
- Generation status
- Current
- Last verified
- 03 Sept 2026
Data & Evidence
Public source evidence is not currently available for Kovara-tracked coverage.