Jake Tech Lab

NVIDIA H100 vs AMD MI355X

AMD MI355X comes out ahead of NVIDIA H100 on 7 of the 8 things that decide this, going by what each maker publishes. The one that matters for you may be one the leader loses.

NVIDIA H100
NVIDIA H100
6.0
Fair
1 of 8 rounds
vs
AMD MI355XWinner
AMD MI355X
7.3
Good
7 of 8 rounds

The verdict

AMD MI355X takes it, 7–1 on rounds.

On the published figures, AMD MI355X takes 7 of 8 rounds. Treat that as a starting point and check the current figures on both pages, because prices and plans move.

Choose NVIDIA H100 if…

  • • You want NVIDIA H100's edge on how much power it draws

Choose AMD MI355X if…

  • • You want AMD MI355X's edge on how much memory it has and how fast its memory is

Round by round

  1. Round 1How much memory it hasAMD MI355X wins
    NVIDIA H1005/10

    80 GB HBM3

    gpu memory

    • 80 GB HBM3, against 288 GB HBM3E

    Source: nvidia.com ↗

    AMD MI355X8/10

    288 GB HBM3E

    dedicated memory size

    Source: amd.com ↗

    AMD MI355X wins this round on what the company itself publishes.

  2. Round 2How fast its memory isAMD MI355X wins
    NVIDIA H1005/10

    3.35 TB/s

    memory bandwidth

    • 3.35 TB/s, against 8 TB/s

    Source: nvidia.com ↗

    AMD MI355X8/10

    8 TB/s

    peak memory bandwidth

    Source: amd.com ↗

    AMD MI355X wins this round on what the company itself publishes.

  3. Round 3How fast it runs FP8 AI mathAMD MI355X wins
    NVIDIA H1005/10

    1,979 TFLOPS

    fp8 tensor core

    • 1,979 TFLOPS, against 5 PFLOPs

    Source: dam-cdn.nvd.orangelogic.com ↗

    AMD MI355X8/10

    5 PFLOPs

    peak ocp-fp8 matrix

    Source: amd.com ↗

    AMD MI355X wins this round on what the company itself publishes.

  4. Round 4How fast it runs FP16 AI mathAMD MI355X wins
    NVIDIA H1005/10

    989 TFLOPS

    fp16 tensor core

    • 989 TFLOPS, against 2.5 PFLOPs

    Source: dam-cdn.nvd.orangelogic.com ↗

    AMD MI355X8/10

    2.5 PFLOPs

    peak fp16 matrix

    Source: amd.com ↗

    AMD MI355X wins this round on what the company itself publishes.

  5. Round 5How much power it drawsNVIDIA H100 wins
    NVIDIA H1008/10

    700 W

    max tdp

    Source: nvidia.com ↗

    AMD MI355X5/10

    1400W

    typical board power

    • 1400W, against 700 W

    Source: amd.com ↗

    NVIDIA H100 wins this round on what the company itself publishes.

  6. Round 6How it handles scientific FP64 mathAMD MI355X wins
    NVIDIA H1005/10

    67 TFLOPS

    fp64 tensor core

    • 67 TFLOPS, against 78.6 TFLOPs

    Source: nvidia.com ↗

    AMD MI355X8/10

    78.6 TFLOPs

    peak fp64 matrix

    Source: amd.com ↗

    AMD MI355X wins this round on what the company itself publishes.

  7. Round 7Which memory generation it usesAMD MI355X wins
    NVIDIA H1005/10

    HBM3

    memory type

    • HBM3, against HBM3E

    Source: dam-cdn.nvd.orangelogic.com ↗

    AMD MI355X8/10

    HBM3E

    dedicated memory type

    Source: amd.com ↗

    AMD MI355X wins this round on what the company itself publishes.

  8. Round 8How new it isAMD MI355X wins
    NVIDIA H1005/10

    2022

    year announced

    • 2022, against 6/12/2025

    Source: nvidianews.nvidia.com ↗

    AMD MI355X8/10

    6/12/2025

    launch date

    Source: amd.com ↗

    AMD MI355X wins this round on what the company itself publishes.

NVIDIA H100NVIDIA H100
AMD MI355XAMD MI355X

Side by side

AMD MI355XBest for how fast its memory isAMD MI355X
Overall score
6.0
7.3
Best
Price
See price
See price
How fast its memory is
5.7
7.9
How much memory it has
5.4
8.0
How much power it draws
7.1
5.0
How fast it runs FP16 AI math
5.9
8.0
How new it is
5.4
7.8
Head-to-head findings
How fast its memory is
3.35 TB/s
8 TB/s
How much memory it has
80 GB HBM3
288 GB HBM3E
How much power it draws
700 W
1400W
How fast it runs FP16 AI math
989 TFLOPS
2.5 PFLOPs
How new it is
2022
6/12/2025
Which memory generation it uses
HBM3
HBM3E
How fast it runs FP8 AI math
1,979 TFLOPS
5 PFLOPs
How it handles scientific FP64 math
67 TFLOPS
78.6 TFLOPs
Record
Head-to-head record
2–7
7–0
Head-to-heads won
22 %
100 %
Rounds won
31 %
69 %
Pros
      Cons
      • 80 GB HBM3, against 288 GB HBM3E
      • 3.35 TB/s, against 8 TB/s
      • 1,979 TFLOPS, against 5 PFLOPs
      • 989 TFLOPS, against 2.5 PFLOPs
      • 1400W, against 700 W
      • 1400W, against 750W
      • 78.6 TFLOPs, against 163.4 TFLOPs
      • 1400W, against 1000W
      Add a third option to this comparison →

      More AI accelerators match-ups