All AI inference chips

MI300X vs H100

Spec-sheet and pricing comparison of AMD Instinct MI300X and NVIDIA H100 SXM with links to continuously measured LLM inference benchmarks on identical workloads.

Spec-sheet comparison

MI300XH100MI300X / H100
Memory per chip192 GB HBM380 GB HBM32.4x
Memory bandwidth5.3 TB/s3.35 TB/s1.58x
Dense FP8 compute2,615 TFLOP/s1,979 TFLOP/s1.32x
Dense FP4 computeNot supportedNot supportedn/a
TDP750 W700 W1.07x
Hourly rate (neocloud tier)$1.16/hr$1.55/hr0.75x
Scale-up world size8 chips8 chips1.0x

Ratios are spec-sheet values; see the live compare pages for measured deltas.

Frequently asked questions

Which has more memory, MI300X or H100?
MI300X offers 192 GB HBM3 per chip versus 80 GB HBM3 on H100 (2.4x the capacity).
How do MI300X and H100 prices compare?
At the neocloud tier the SemiAnalysis TCO model rates MI300X at $1.16/hr versus $1.55/hr for H100. Hourly price alone is misleading; the per-dollar compare pages divide measured throughput by these rates.
Is MI300X faster than H100 for LLM inference?
On paper MI300X has 1.32x the dense FP8 compute of H100, but delivered tokens per second depend on the model, framework, precision and interactivity target. InferenceX measures both chips daily on identical workloads; see the live compare pages for current results.

See live benchmark results

Every number above is static hardware data. Delivered tokens per second, cost per million tokens and energy per token are measured continuously on the dashboard:

Go deeper with the SemiAnalysis models

InferenceX measures delivered inference performance. The SemiAnalysis institutional models cover the market behind these chips: who ships them, who buys them, and what they cost to own.