AMD Instinct MI325X - 256GB CDNA3 for Inference
AMD Instinct MI325X specs, benchmarks, and analysis. 256GB HBM3e at 6 TB/s, 2.6 PFLOPS FP8, CDNA3 architecture - the memory-capacity upgrade to the MI300X targeting large model inference.
Hbm3e
Huawei Atlas 350 - China's FP4 Inference Accelerator
Huawei Atlas 350 specs, benchmarks, and analysis. Ascend 950PR chip, 112GB HiBL 1.0 HBM, 1.56 PFLOPS FP4, 600W - China's first domestically developed FP4-capable AI accelerator.
Microsoft Maia 200 - Azure's Inference Accelerator
Microsoft Maia 200 specs, benchmarks, and architecture analysis. TSMC 3nm, 216GB HBM3e, 10 PFLOPS FP4, 750W - Microsoft's first inference-only silicon deployed in Azure.
AMD Instinct MI455X
AMD's flagship CDNA 4 AI GPU with 432 GB HBM4, 40 PFLOPS FP4, and 2nm chiplet design targeting H2 2026.
NVIDIA Vera Rubin NVL144
NVIDIA's Rubin-based rack system with 144 R200 GPUs, 3.6 ExaFLOPS FP4, 20 TB HBM4 - arriving H2 2026.
AWS Trainium3 - Amazon's 3nm AI Accelerator
Complete specs, benchmarks, and analysis of AWS Trainium3 - Amazon's TSMC 3nm AI chip with 2.52 PFLOPS FP8, 144GB HBM3e, and NeuronLink-v4, powering Anthropic's Claude through Project Rainier.
Etched Sohu - Transformer-Only Inference ASIC
Full specs and critical analysis of the Etched Sohu - a transformer-specific ASIC claiming 500K+ tokens/sec on Llama 70B, built on TSMC 4nm with 144GB HBM3E. Bold claims, but no independent benchmarks yet.
AMD Instinct MI440X - CDNA 5 Enterprise GPU
Complete specs and analysis of the AMD Instinct MI440X - a CDNA 5 enterprise GPU with 432GB HBM4, 19.6 TB/s bandwidth, and TSMC 2nm compute chiplets.
NVIDIA Rubin R200 - Next-Gen AI Superchip
Complete specs, benchmarks, and analysis of the NVIDIA Rubin R200 GPU - the post-Blackwell flagship with 288GB HBM4, 22 TB/s bandwidth, and 50 PFLOPS FP4.
AMD Instinct MI300X
AMD Instinct MI300X specs, benchmarks, and real-world performance data. 192GB HBM3, 5,300 GB/s bandwidth, 2,610 TFLOPS FP8 on CDNA 3 chiplet architecture.
AMD Instinct MI350X
AMD Instinct MI350X specs and performance estimates. 288GB HBM3e, ~6,000 GB/s bandwidth, ~3,600 TFLOPS FP8 on CDNA 4 architecture at TSMC 3nm.
AWS Trainium2 - Amazon's Cloud Training Chip
AWS Trainium2 is Amazon's second-generation custom AI training chip, powering EC2 Trn2 instances with 96GB HBM2e per chip and tight integration with the AWS Neuron SDK and SageMaker ecosystem.
