Uptime Institute's 16th Annual Global Data Center Survey reports modal rack density reached 11 kW in 2026, up from 9 kW in 2025, while AI racks such as Nvidia's GB300 NVL72 require up to 142 kW per ra…
On June 12, 2026, Artificial Analysis, in collaboration with NVIDIA, released AA-AgentPerf, a benchmark that introduces 'Agents per Megawatt' as a primary metric for agentic AI hardware performance. I…
Alibaba released the model weights for Qwen3.8-Flash-Next, a multimodal mixture-of-experts model with 176B total parameters, as a preview of the upcoming Qwen4 architecture. NVIDIA provides Day 0 func…
NVIDIA Corp. detailed its Rubin GPU and Vera Rubin NVL72 rack at Hot Chips 2026, targeting agentic AI workloads with a full-stack platform spanning seven chips and five racks. The company claims up to…
NVIDIA introduced NVLink Fusion, a technology that connects custom XPUs to NVIDIA's AI infrastructure, aiming to reduce cost and time-to-market for AI factory builders. The solution offers 3x lower la…
NVIDIA's Vera Rubin NVL72 systems deliver up to 30x higher throughput per megawatt than the GB300 NVL72 on agentic AI workloads, according to new measured performance data from NVIDIA using the SemiAn…
SemiAnalysis's AgentX benchmark shows NVIDIA's Vera Rubin NVL72 delivers up to 30x higher AI-factory throughput per megawatt than GB300 NVL72 for agentic-coding inference, while GB300 NVL72 extends it…
HIVE Digital Technologies announced a five-year GPU cloud services agreement worth approximately $350 million through its subsidiary BUZZ HPC during its Fiscal Q1 2027 earnings call on August 17. The …
Alibaba's Qwen team launched Qwen3.8-Max on August 3, a 2.4 trillion parameter sparse Mixture-of-Experts model that runs at over 4,000 tokens per second per GPU on Nvidia's GB300 NVL72 hardware, scori…
Alibaba released the open weights for Qwen3.8-2.4T-A95B (Qwen3.8-Max), a 2.4T-parameter mixture-of-experts model with 95B activated parameters per token, designed for long-context reasoning and agenti…
Cursor Research released Mixture-of-Kittens (MoK), a CUDA kernel that fuses mixture-of-experts computation and communication, delivering a 41% training throughput increase on Nvidia GB300 NVL72 system…
Cursor, the AI coding company, open-sourced Mixture-of-Kittens (MoK), a deterministic MoE training megakernel for NVL72s that fuses communication and computation into a single kernel, achieving a 1.41…
NVIDIA Dynamo, an Apache-2.0 datacenter-scale inference orchestration layer from NVIDIA, sits above vLLM, SGLang, and TensorRT-LLM to coordinate multi-GPU clusters, with published benchmarks showing u…
Nvidia's Jensen Huang announced at GTC that Vera Rubin delivers approximately 10x more inference throughput per watt compared to Blackwell (B200), but the comparison point matters because most organiz…
VLLM announces efficient day-0 support for Moonshot AI's Kimi K3, a 2.8-trillion-parameter Mixture-of-Experts model, achieving up to 370 tokens per second with speculative decoding on 16 NVIDIA GB300 …
NVIDIA GB300 NVL72 set a world record for pre-training DeepSeek-V3 671B at 1,648 TFLOPs per GPU, demonstrating how advances across the AI platform—from silicon to networking to software—push training …
NVIDIA's GB300 NVL72 platform delivers up to 25x performance per watt over its predecessor, the NVIDIA Hopper, redefining AI infrastructure efficiency. The company's extreme codesign from silicon to s…
NVIDIA's Blackwell platform swept MLPerf Training 6.0, achieving the fastest training times across all seven benchmarks, scaling to 8,192 GPUs on DeepSeek-V3 671B, and delivering up to 1.6x performanc…
Artificial Analysis released AgentPerf, the first benchmark for agentic-AI infrastructure, which replays recorded multi-step agent trajectories instead of single chat completions. NVIDIA reported that…
Nvidia's new GB300 NVL72 system achieves 61,400 concurrent AI agents per megawatt, a 20x improvement over the prior-generation H200. The Blackwell Ultra rack-scale system, validated using the AgentPer…