Specifications Compared
| Spec | Gaudi 2 | MI300X |
|---|---|---|
| TDP | 600W | 750W |
| VRAM | 96 GB | 192 GB |
| FP8 (dense) | Not published | 2,614.9 TFLOPS |
| Memory Type | HBM2e | HBM3 |
| Architecture | Gaudi | CDNA 3 |
| FP16 (dense) | Not published | 1,307.4 TFLOPS |
| Form Factors | OAM | OAM |
| INT8 (dense) | Not published | 2,614.9 TOPS |
| Interconnect | RoCE Ethernet, PCIe 4.0 | Infinity Fabric, PCIe 5.0 |
| FP32 Performance | Not published | 163.4 TFLOPS |
| FP64 Performance | Not published | 81.7 TFLOPS |
| Memory Bandwidth | 2,450 GB/s | 5,300 GB/s |
| FP8 (with sparsity) | Not published | 5,229.8 TFLOPS |
| FP16 (with sparsity) | Not published | 2,614.9 TFLOPS |
| INT8 (with sparsity) | Not published | 5,229.8 TOPS |
Performance Analysis
Memory bandwidth differences shape achievable batch sizes during model execution. The MI300X supplies 5300 GB/s bandwidth while the Gaudi 2 supplies 2450 GB/s bandwidth so the MI300X supports larger batches before memory stalls occur. The MI300X publishes a dense FP16 rating of 1307.4 TFLOPS and a rating with sparsity of 2614.9 TFLOPS together with a dense FP32 rating of 163.4 TFLOPS. Because the Gaudi 2 lists no tensor figures the comparison of dense FP16 against dense FP16 remains limited to the published MI300X values. Higher bandwidth on the MI300X also pairs with its 192 GB capacity to accommodate models that exceed the 96 GB limit of the Gaudi 2.
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.
Gaudi 2
Gaudi 2 is not offered on-demand by any provider we track right now. See the Gaudi 2 rental page for last-seen listed prices and a price alert.
MI300X
MI300X is not offered on-demand by any provider we track right now. See the MI300X rental page for last-seen listed prices and a price alert.
Notify me when Gaudi 2 drops below a price
One email when the cheapest in-stock on-demand price per GPU-hour falls below your threshold.
QuantaCloud
Comparing providers? We broker across all of them.
Stop tab-switching between pricing pages. Tell us what you need, 16+ GPUs reserved or cluster capacity, and we return one quote at partner rates within 24 hours.
When to Choose the Gaudi 2
The Gaudi 2 suits deployments where a 600W TDP is required to meet facility power limits. Its RoCE Ethernet interconnect matches existing Ethernet fabrics without requiring Infinity Fabric adapters.
When to Choose the MI300X
The MI300X suits workloads that need 192 GB capacity and 5300 GB/s bandwidth to hold larger models or increase batch sizes. Its published dense FP16 rating of 1307.4 TFLOPS and rating with sparsity of 2614.9 TFLOPS provide measurable throughput targets for training and inference.
Use Cases
The MI300X supplies 192 GB VRAM and 5300 GB/s bandwidth against 96 GB and 2450 GB/s on the Gaudi 2 allowing larger models to fit without partitioning.
The MI300X lists a dense FP16 rating of 1307.4 TFLOPS and 192 GB capacity that together support higher throughput than the Gaudi 2 can demonstrate with no published tensor figures.
The MI300X provides 5300 GB/s bandwidth and a rating with sparsity of 2614.9 TFLOPS for FP16 that accelerate gradient updates compared with the lower bandwidth of the Gaudi 2.
Both devices share the OAM form factor and the MI300X 192 GB capacity can be traded against the Gaudi 2 600W TDP depending on power constraints.
The MI300X publishes a dense FP32 rating of 163.4 TFLOPS while the Gaudi 2 lists no FP32 figure and the higher 5300 GB/s bandwidth aids large matrix operations.
Frequently Asked Questions
What is the memory bandwidth of the MI300X?▾
The MI300X supplies 5300 GB/s bandwidth. This figure exceeds the 2450 GB/s bandwidth listed for the Gaudi 2.
Does the Gaudi 2 publish FP16 performance?▾
The Gaudi 2 does not publish an FP16 figure. The MI300X publishes a dense FP16 rating of 1307.4 TFLOPS and a rating with sparsity of 2614.9 TFLOPS.
What interconnect does the MI300X support?▾
The MI300X supports Infinity Fabric and PCIe 5.0. The Gaudi 2 supports RoCE Ethernet and PCIe 4.0.
How does TDP differ between the two devices?▾
The Gaudi 2 lists a 600W TDP while the MI300X lists a 750W TDP. Both devices use the OAM form factor.
What FP32 rating does the MI300X publish?▾
The MI300X publishes a dense FP32 rating of 163.4 TFLOPS. The Gaudi 2 lists no FP32 figure.
Which is cheaper to rent, the Gaudi 2 or the MI300X?▾
Cloud rental prices for both the Gaudi 2 and MI300X 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 Gaudi 2 have compared to the MI300X?▾
The Gaudi 2 has 96 GB of HBM2e memory. The MI300X has 192 GB of HBM3 memory.
Can I find Gaudi 2 and MI300X 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 Gaudi 2 and the MI300X?▾
The Gaudi 2 uses the Gaudi architecture (2022) while the MI300X uses CDNA 3 (2023). The Gaudi 2 has 96 GB of HBM2e at 2,450 GB/s; the MI300X has 192 GB of HBM3 at 5,300 GB/s, so the MI300X has 2.2x the memory bandwidth of the Gaudi 2.
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 Gaudi 2 and the MI300X. 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 GAUDI2Every current offer by provider, daily price history and a price alert.
- Rent MI300XEvery 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.