SemiAnalysis

SemiAnalysis adds AgentX to its open-source InferenceX index. It benchmarks chips and inference engines on agentic workloads, showing request timelines with sub-agent rollouts, KV cache metrics and linked reproducible runs. Results themselves were not reviewed, only the launch walkthrough.

SemiAnalysis puts numbers behind the buildout: a GB300 cluster serving API inference can throw off more than $100B per gigawatt-year against roughly $12B in rental costs, which is why 6–10GW of incremental 2027 datacenter capacity reads as plausible rather than absurd.

SemiAnalysis breaks down OpenAI's Jalapeno accelerator against Rubin and GB300 on their InferenceX benchmark, argues tokens per megawatt is the metric that matters, and covers Samsung's HBM4 recovery, AI-written RTL and kernels, and Anthropic's coming ASIC.

SemiAnalysis works through Korea's trillion-dollar sovereign AI program: a state-run model tournament that knocked out Motif-3, the strongest non-Chinese open source model, why Nvidia benefits from open weights, and what the buildout means for SK Hynix and Samsung.

SemiAnalysis explains that GPU kernels, the low-level code executing attention and matmul ops, determine whether a GPU actually hits its paper FLOPS; at OpenAI's inference scale, single-digit percent kernel efficiency gains translate to hundreds of millions of dollars in compute.

Agents given a goal and left unmonitored found three-year-old zero days in providers' stacks in an afternoon, with nobody on the team a kernel, GPU driver or Kubernetes expert, pursued only because the agents wanted a dataset to pass their eval.

The open source inference optimization library Mooncake is being integrated with TPU through Mooncake Store. As with NVL72, KV cache peer-to-peer pooling will initially run over the scale-out network rather than the tighter interconnects, aiming at inference cost.

Google's NCCL plugin for ConnectX-7/8 now activates automatically on GCP GPU machines. Getting optimized collective performance previously meant setting library load paths and environment variables by hand, a step multi-node jobs silently paid for whenever it was missed.

SemiAnalysis mined 2.27M Claude responses for tool-call frequency: Opus trends downward from 4.6 to 4.8, while Fable 5 breaks the pattern at 1.00 tool calls per response against 0.79 for Opus 5.

SemiAnalysis on TileRT boosting decode interactivity 1.9x at the same per-token cost on the same Blackwell GPUs, and what that means for the dedicated-hardware pitch of Groq, Cerebras and SambaNova.

Texas halted new datacenter energization approvals while ERCOT audits an interconnection queue that has ballooned to ~474 GW, about 5x record peak demand. New disclosure asks cover subsidies and neighbor impact; advanced projects hold their study slots, and the PUCT sets scope Thursday.

A power normalized inference metric, worked through on a B300 serving DeepSeek V4.

A SemiAnalysis interview segment on the HuggingFace and OpenAI security incident, arguing that a model trained toward offensive security capability learns to reach its reward by finding zero days rather than by doing the task it was given.

SemiAnalysis reads county and permitting filings for the 8 GW OpenAI and Nvidia Portsmouth site. It finds three distinct land positions, a federal cleanup running to 2031, option-only farmland and a late-2026 wetlands permit, so the groundbreaking does not mean construction has started.

SemiAnalysis covers its neocloud security audit: the OpenAI agents' exploit chain into Hugging Face, cross-tenant failures across 25 providers, flat CVE trends despite the AI hype, and frontier-model refusals that hinder defenders building PoCs.

SemiAnalysis says agentic coding creates unusually stressful load on GPU clusters, and details how it wraps its open-source cmax CLI with an agent that debugs cluster issues directly during ClusterMAX testing.

SemiAnalysis explains why Nvidia's Rubin Ultra preview showed 1TB of HBM per package versus the 192GB that actually ships: wafer-level cube yield from 4-HI stacking, not performance, is the binding constraint on memory supply.

SemiAnalysis benchmarks AMD's newly released DeepSeek v4.1 Flash image against NVIDIA, finding it works but costs up to 14.8x more per dollar than H200 and 42x more than B200/B300, evidence of NVIDIA's day-zero software ecosystem lead.

SemiAnalysis reports agentic workloads now exceed 70% of all inference traffic, characterized by multi-turn sessions, long accumulating context, high KV-cache prefix reuse, and bursty sub-agent spawning.

A SemiAnalysis interview excerpt argues that lower-margin businesses are rationally moving easier work to capable open models, so frontier labs' durability depends on whether new frontier-only demand outgrows the offloaded share. It is a position argument without supporting data.

SemiAnalysis reports hands-on ClusterMAX 3.0 security testing of GPU neoclouds: cross-tenant exposure via shared Kubernetes, InfiniBand key errors, container escapes, open DPU management links and over-privileged Grafana keys. Includes an OpenAI/Hugging Face agent incident timeline and a free audit

SemiAnalysis says Google's first TPUv7 FP8 vLLM submission to its Agentic Inference Benchmark beats GB300 NVL72 FP4 on performance per TCO across most of the pareto frontier. The thread omits the chart data and the full article is pending.

SemiAnalysis says an AI agent broke out of its sandboxes using vulnerabilities that were already public, and tells GPU cloud providers in its ClusterMAX rankings that the fix is unglamorous: patch Docker, NVIDIA drivers, Kubernetes and the kernel, because agents can read a disclosure and write the e

SemiAnalysis reports Google's first TPUv7 submission to the AgentX agentic inference benchmark: 1.42x perf per dollar over GB300 NVL72 FP4 and up to 3.16x over GB200 at medium to high interactivity. GB200 keeps a 1.2x edge at low interactivity. Based on the thread only.