Specifications Compared
| Spec | A40 | B300 |
|---|---|---|
| TDP | 300W | 1400W |
| VRAM | 48 GB | 262-288 GB |
| CUDA Cores | 10,752 | Not published |
| FP4 (dense) | Not published | 13,500 TFLOPS |
| FP8 (dense) | Not published | 4,500 TFLOPS |
| Memory Type | GDDR6 | HBM3e |
| Architecture | Ampere | Blackwell Ultra |
| FP16 (dense) | 149.7 TFLOPS | 2,250 TFLOPS |
| Form Factors | PCIe | SXM6 |
| INT8 (dense) | 299.3 TOPS | 187.5 TOPS |
| Interconnect | NVLink, PCIe 4.0 | NVLink, PCIe 6.0, InfiniBand |
| Tensor Cores | 336 | Not published |
| FP32 Performance | 37.4 TFLOPS | 75 TFLOPS |
| FP64 Performance | Not published | 1.25 TFLOPS |
| Memory Bandwidth | 696 GB/s | 8,000 GB/s |
| FP4 (with sparsity) | Not published | 18,000 TFLOPS |
| FP8 (with sparsity) | Not published | 9,000 TFLOPS |
| FP16 (with sparsity) | 299.4 TFLOPS | 4,500 TFLOPS |
| INT8 (with sparsity) | 598.6 TOPS | 375 TOPS |
Performance Analysis
The FP16 dense performance of the B300 at 2250 TFLOPS exceeds the FP16 dense performance of the A40 at 149.7 TFLOPS by a factor of fifteen. This gap means training sessions that rely on dense FP16 arithmetic complete faster on the B300 while inference pipelines using the same dense FP16 arithmetic also finish more operations per second on the B300. Memory bandwidth of 8000 GB per second on the B300 compared with 696 GB per second on the A40 permits larger batch sizes during both training and inference because data movement keeps pace with the higher arithmetic throughput. When sparsity is enabled the B300 reaches 4500 TFLOPS in FP16 with sparsity against 299.4 TFLOPS in FP16 with sparsity on the A40 so workloads that exploit sparsity see the same proportional advantage on the newer GPU.
Current On-Demand Offers
Cheapest secure on-demand offer per provider that is reported in stock right now, per GPU per hour; multi-GPU instances show the whole-instance price alongside. Deploy opens the DeployGPU console for providers on the platform, otherwise the provider's own site. Stock is rechecked every minute.
A40
| Provider | Region | GPUs | Per GPU / hr | Instance / hr | Deploy |
|---|---|---|---|---|---|
| LeaderGPU | The Netherlands | 8 | $0.52 | $4.13 | Deploy |
| RunPod | global | 1 | $0.59 | — | Deploy |
2 providers in stock, 2 offers. All A40 offers, price history and alerts
B300
1 provider in stock, 2 offers (cheapest per provider shown). All B300 offers, price history and alerts
Notify me when A40 drops below a price
One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold. Cheapest right now: $0.52/GPU-hr.
QuantaCloud
Comparing B-series options? Get one quote for all of them.
Skip the per-provider sales calls. Reserved and cluster B-series configurations from 16 to 1024+ GPUs with InfiniBand fabric, 3 to 12 month terms. One quote at partner rates, 24h turnaround.
When to Choose the A40
The A40 suits deployments where power draw must stay near 300 W TDP and where 48 GB of GDDR6 memory meets capacity needs. Workloads that fit within 696 GB per second bandwidth and require only 37.4 TFLOPS in FP32 operate efficiently on this PCIe card without exceeding thermal or electrical limits of smaller systems.
When to Choose the B300
Systems that accommodate 1400 W TDP and SXM6 form factors gain from the 262 GB HBM3e capacity when models exceed the limits of smaller memory pools.
Use Cases
The B300 supplies 2250 TFLOPS in dense FP16 against 149.7 TFLOPS in dense FP16 on the A40 so training runs finish with fewer devices.
The B300 supplies 2250 TFLOPS in dense FP16 and 8000 GB per second bandwidth which allow larger batch sizes than the A40 can sustain.
The B300 reaches 4500 TFLOPS in FP16 with sparsity while the A40 reaches 299.4 TFLOPS in FP16 with sparsity so adaptation steps complete faster.
The A40 meets typical image generation memory needs at 48 GB while consuming only 300 W TDP in environments that cannot support 1400 W.
The A40 delivers 37.4 TFLOPS in FP32 at 300 W TDP for simulations that do not require the higher 75 TFLOPS in FP32 of the B300.
Frequently Asked Questions
What is the memory difference between the A40 and the B300?▾
The A40 contains 48 GB of GDDR6 memory while the B300 contains between 262 GB and 288 GB of HBM3e memory.
How does FP32 performance compare on the A40 versus the B300?▾
The A40 provides 37.4 TFLOPS in FP32 and the B300 provides 75 TFLOPS in FP32.
Which GPU offers higher dense FP16 throughput?▾
The B300 offers 2250 TFLOPS in dense FP16 while the A40 offers 149.7 TFLOPS in dense FP16.
What interconnect options exist on each card?▾
The A40 supports NVLink and PCIe 4.0 while the B300 supports NVLink, PCIe 6.0, and InfiniBand.
How do the TDP ratings differ?▾
The A40 has a TDP of 300 W and the B300 has a TDP of 1400 W.
Which architecture does each GPU use?▾
The A40 uses the Ampere architecture and the B300 uses the Blackwell Ultra architecture.
Which is cheaper to rent, the A40 or the B300?▾
Cloud rental prices for both the A40 and B300 vary by provider, configuration, and availability. This page shows live pricing from 25+ providers updated every 60 seconds. Scroll to the Live Cloud Pricing section to compare current rates.
How much VRAM does the A40 have compared to the B300?▾
The A40 has 48 GB of GDDR6 memory. The B300 has 262 to 288 GB of HBM3e memory.
Can I find A40 and B300 GPUs available to rent right now?▾
Yes. This page shows real-time availability across 25+ cloud GPU providers. The Live Cloud Pricing section displays only in-stock offers with current pricing.
What is the main difference between the A40 and the B300?▾
The A40 uses the Ampere architecture (2020) while the B300 uses Blackwell Ultra (2025). The B300 delivers 15.0x the dense FP16 throughput (2,250 vs 149.7 TFLOPS, both without sparsity) and 11.5x the memory bandwidth of the A40.
Rent these GPUs
Each GPU page lists every current on-demand offer by provider, per GPU-hour, updated every minute.
Related comparisons
How this page is made
- Specifications come from the NVIDIA, AMD and Intel datasheets for the A40 and the B300. Dense and sparse throughput are listed separately.
- Prices and availability are live: every offer shown is an in-stock, on-demand listing from a provider we track, rechecked every minute.
- The written comparison (overview, performance notes, when to choose each card, verdict, use cases and FAQ) was drafted with an AI model from the specification table above and passed an automated check that rejects any figure not in that table. It was last generated on . It contains no prices; those are always read live.
- Read how we collect the data, or report an error on this page.
Next steps
- Rent A40Every current offer by provider, daily price history and a price alert.
- Rent B300Every current offer by provider, daily price history and a price alert.
- GPU price indexHow on-demand prices for the major GPUs have moved, updated daily.
- All GPUsEvery GPU we track, with current lows and provider counts.