NVIDIA B300 SXM vs NVIDIA RTX 6000 Ada 48GB
B300 SXM has 500% more memory than RTX 6000 Ada 48GB (288 GB vs 48 GB). Its memory bandwidth is 733% more (8 vs 0.96 TB/s). In the cloud, the median on-demand B300 SXM price is $7.64 per GPU-hour against $0.79 for the RTX 6000 Ada 48GB. Per gigabyte of GPU memory, that is $0.027 vs $0.016 per hour.
Specifications
Manufacturer figures as recorded by Kovara. Higher memory and bandwidth are highlighted; neither is a benchmark of your workload.
| B300 SXM | RTX 6000 Ada 48GB | |
|---|---|---|
| Manufacturer | NVIDIA | NVIDIA |
| Architecture | Blackwell Ultra | Ada Lovelace |
| Memory | 288 GB HBM3E | 48 GB GDDR6 |
| Memory bandwidth | 8 TB/s | 0.96 TB/s |
| Power (TDP) | 1400 W | 300 W |
| Form factor | SXM | PCIe dual-slot |
Cloud prices today
Public offers Kovara tracks, normalised to US dollars per GPU-hour. Prices use on-demand terms where available; spot and reserved offers are on each GPU’s price pages. Lower is highlighted.
| B300 SXM | RTX 6000 Ada 48GB | |
|---|---|---|
| Lowest price | $6.60 | $0.79 |
| Median price | $7.64 | $0.79 |
| Highest price | $12.66 | $1.64 |
| Median per GB of memory | $0.027 | $0.016 |
| Public offers (all terms) | 30 | 9 |
| Providers | 12 | 4 |
Where to rent each
Lowest tracked price per provider. Open a provider to see every offer, region and term.
Questions
Which is cheaper to rent, the B300 SXM or the RTX 6000 Ada 48GB?
Right now the cheapest on-demand B300 SXM Kovara tracks is $6.60 per GPU-hour and the cheapest RTX 6000 Ada 48GB is $0.79. Medians are $7.64 and $0.79. The cheaper GPU per hour is not always cheaper per job: a faster or larger GPU can finish sooner or need fewer units.
Which has more memory?
The B300 SXM has more memory: 288 GB against 48 GB. Memory decides which models and batch sizes fit on one GPU.
Are these benchmark results?
No. This page compares manufacturer specifications and public prices. Real performance depends on the model, precision, software and interconnect. Kova can help you estimate cost for a specific workload.