MI400 vs B300 Blackwell Ultra vs B200 vs Ascend 950DT

Pyyan / Compare / MI400 vs B300 Blackwell Ultra vs B200 vs Ascend 950DT

4 of 5

GPU & AI Chips · verified 24 Sept 2026

×MI400AMDcurrent

×B300 Blackwell UltraNVIDIAcurrent

×B200NVIDIAcurrent

×HAscend 950DTHuaweicurrent

Specification	MI400	B300 Blackwell Ultra	B200	Ascend 950DT
Summary	Double NVIDIA's memory capacity.	15 PFLOPS dense FP4.	The current volume Blackwell part.	144GB on Huawei's own HBM, and the part China is building gigawatts around.
Memory	432GB HBM4	288GB HBM3e	192GB HBM3e	144GB HiZQ 2.0
Bandwidth	19.6 TB/s	8 TB/s	—	4 TB/s
Price	—	~$40–50K (verify)	$30–50K	Above ¥250,000 (about $37K), reported
Vendor	AMD	NVIDIA	NVIDIA	Huawei
Category	GPU & AI Chips	GPU & AI Chips	GPU & AI Chips	GPU & AI Chips
Official	AMD ↗	NVIDIA ↗	NVIDIA ↗	Huawei ↗

Highlighted rows are where these differ.

Best for large-model inference.

Best for frontier training.

Best for general training.

4 TB/s memory bandwidth and a 2 TB/s interconnect, 2.5x the Ascend 910C
HiZQ 2.0 is Huawei's own memory standard rather than a licensed HBM3 or HBM3e part
Indicated price rose 20% to 50% in two months to September 2026, on a memory shortage
Successor Ascend 960DT pulled forward to Q1 2027, Huawei said on 17 September

Best for serving models where Nvidia cannot be bought.