Skip to main content
newspals
Topics
Concepts
Editors
Newsletter
English
SemiAnalysis — Concepts | NewsPals
Concepts
·
SemiAnalysis
the lore behind the feed
SemiAnalysis
The stories that keep pulling this idea back into the feed.
5 stories
In the feed
cybersecurity
SemiAnalysis says most neoclouds make OpenAI vs HuggingFace scale AI infrastructure a supply chain risk
AI compute buyers need to treat rented GPU capacity as a third-party trust boundary, not just a procurement win.
ai-ml
SemiAnalysis on OpenAI Jalapeño: why labs stop renting GPUs
The spicy bit is not the chip name. It is the math of serving tokens when inference becomes the product.
ai-ml
SemiAnalysis AgentX InferenceXv3 makes KV cache the agent inference battleground
Agent workloads are not just longer chats. They are messy, multi-turn cost machines where serving behavior now matters as much as model choice.
ai-ml
SemiAnalysis on Meta’s post-Llama 4 AI operating model: $14.3B Scale AI, 2000km+ scale-across, and the Tent
Meta’s superintelligence rebuild is less about slogans and more about talent concentration, compute plumbing, and release strategy.
hardware
SMIC N+3 Teardown: How DUV Multi-Patterning Outpitched Intel 18A
SemiAnalysis reverse-engineered the Kirin 9030 and found a 32.5 nm minimum metal pitch. Here is what that number teaches you about process technology.
Also vibing
AI Infrastructure
Agentic AI
AgentX InferenceXv3
AI Datacenters
Broadcom
DUV Lithography
Huawei Kirin 9030
Inference