NVIDIA Groq 3 LPU - SRAM-Based Inference Engine
The NVIDIA Groq 3 LPU is a pure-SRAM inference chip delivering 150 TB/s memory bandwidth and 1.2 PFLOPS FP8 per chip, designed to pair with Vera Rubin GPUs for trillion-parameter model serving.
Asic
Cerebras Launches $2B IPO Roadshow on Nasdaq
Cerebras Systems has kicked off a $2 billion IPO roadshow targeting a Nasdaq listing under ticker CBRS, anchored by a $10 billion compute contract with OpenAI.
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.
Percepta Builds a Computer Inside a Transformer
Percepta AI compiled a WebAssembly interpreter into transformer weights, executing programs deterministically at 33K tokens/sec on CPU - but the community is skeptical about the practical value.
Meta MTIA 300
Meta's first mass-deployed RISC-V AI accelerator - 1.2 PFLOPS FP8, 216 GB HBM, powering Facebook and Instagram at scale.
NVIDIA's Secret Chip Fuses GPU and Groq for OpenAI
NVIDIA will unveil a new inference processor built on Groq's LPU architecture at GTC 2026, with OpenAI as its first major customer allocating 3 GW of dedicated capacity.
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.
Qualcomm AI200 - Rack-Scale Inference ASIC
Qualcomm AI200 specs and analysis - a Hexagon-based inference accelerator with 768GB LPDDR per card, rack-scale design, and a focus on inference TCO.
SambaNova SN50 RDU - Agentic Inference Chip
Complete specs and analysis of SambaNova's SN50 RDU - a TSMC 3nm dataflow chip with 3.2 PFLOPS FP8, three-tier memory, and claimed 5x speed over NVIDIA B200.
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.
Cambricon MLU590 - China's Inference Accelerator
Full specs and analysis of the Cambricon MLU590 - 192GB HBM2e, ~2,400 GB/s bandwidth, TSMC 7nm, and what it means for AI inference outside the NVIDIA ecosystem.
