Hardware & Compute
Chips, data centres and the infrastructure behind AI.
-

Groq LPU vs SambaNova SN40L: Llama 4 inference comparison
SambaNova’s SN40L outperforms Groq’s LPU for Llama 4 Maverick deployments by utilizing a three-tier memory system. While Groq requires 576 chips for…
-

The history of NVIDIA NVLink from Kepler to Blackwell Ultra
Trace the evolution of NVIDIA NVLink from the 160 GB/s Tesla P100 to the 3.6 TB/s Vera Rubin platform. This technical history…
-

Tesla shifts from Dojo to AI6 training clusters
Elon Musk has disbanded the Dojo team in favor of the AI6 chip architecture, which scales from FSD to Optimus. This strategic…
-

Why Cerebras’s Andromeda supercomputer matters for OpenAI
OpenAI has invested $10 billion in Cerebras to address inference bottlenecks for models like GPT-5. The Andromeda supercomputer utilizes wafer architecture to…
-

Lightmatter’s Passage photonic interconnect and NVIDIA Blackwell
Lightmatter’s Passage M1000 photonic interposer provides 114 Tbps of bidirectional bandwidth to solve chiplet communication bottlenecks. This technology integrates with NVIDIA’s Blackwell…
-

Power delivery failures in SambaNova SN40L rack deployments
Deploying SambaNova SN40L racks requires precise power and thermal validation to avoid instability. Common mistakes include insufficient PDU margin testing and overlooking…
-

Risks in Intel’s Falcon Shores delays for Argonne’s Aurora 2 upgrade
Intel’s shifting GPU roadmap and the transition to Falcon Shores create hardware gaps that threaten exascale deployments like Argonne National Laboratory’s Aurora…
-

Samsung Foundry and Tenstorrent build next-generation AI chiplets
Samsung Foundry manufactures Tenstorrent’s RISC-V Wormhole AI chiplets using the SF4X 4nm process. Led by Jim Keller, Tenstorrent aims to provide an…
-

Groq LPU inference pricing edge over NVIDIA H100s
Groq provides a cost advantage for Llama 3.3 70B workloads compared to NVIDIA H100s when using highly optimized hardware. While Cerebras offers…
-

SambaNova SN40L claims vs reality in enterprise LLM deployments
The SambaNova SN40L RDU excels at low-concurrency MoE workloads like Llama 4 Maverick, delivering 2,800 tokens per second. However, NVIDIA H200 and…
-

Physical constraints of AMD MI400 and OpenAI Stargate timeline
AMD faces significant supply chain risks for its MI400 series chips due to TSMC CoWoS packaging bottlenecks and HBM4 memory production limits.…
Trending Now
- The technical evolution of Stable Diffusion
- Risks in Anthropic’s Claude 3.5 Opus enterprise and healthcare deployment
- Runway Gen-4: myths vs facts about commercial licensing
- The transition to agentic workflows and multi-model architectures
- Waymo scales robotaxi operations amid economic and regulatory pressures

