SpaceX to Provide Anthropic Access to AI Supercomputer
SpaceX is set to give Anthropic access to its massive AI supercomputer, expanding the safety lab’s compute resources beyond AWS in a deal between two major U.S. tech players.
-
-
Anthropic Tapped xAI’s Colossus Supercomputer to Fix Claude’s Sycophancy Problem
Anthropic used xAI’s 220,000-GPU Colossus supercomputer to fix Claude’s sycophancy problem — a rare example of compute sharing between rival AI companies. #AI #Anthropic #xAI
-
AMD, OpenAI Partner on AI Networking With MRC
AMD and OpenAI are teaming up on MRC-based AI networking tech, targeting the interconnect layer linking thousands of accelerators — a key Nvidia stronghold. #AI #AIInfrastructure
-
NVIDIA Adds Multi-Rail Connectivity to Spectrum-X Ethernet Fabric
NVIDIA upgrades Spectrum-X Ethernet fabric with Multi-Rail Connectivity, letting GPU servers tap multiple network paths simultaneously — targeting bandwidth bottlenecks in gigascale AI clusters. #AI #Networking #NVIDIA
-
EPAM Partners With Anthropic to Train 20,000 Staff on AI
EPAM Systems partners with Anthropic to train 20,000 employees on Claude AI tools — adding a major global systems integrator to Anthropic’s growing enterprise roster.
-
OpenAI Releases New Network Protocol for AI Training Clusters
OpenAI releases MRC (Multipath Reliable Connection), an open networking protocol for AI training supercomputers, improving GPU cluster resilience via multipath routing during long training runs. #AI #OpenAI
-
Musk Appears to Admit xAI Used OpenAI Models to Train Grok
Elon Musk appears to have acknowledged xAI used OpenAI models to train Grok—amid active litigation between the two AI rivals. The practice of “model distillation” occupies a legal gray area as AI IP law matures.
-
Over 700 Meta AI Training Workers in Ireland Face Layoffs
700+ Meta AI training workers in Ireland face layoffs via contractor Covalen, spotlighting the precarious labor conditions of data annotators powering Big Tech’s AI systems.
