The new Vera Rubin Space Module processed massive satellite datasets directly in space In sum – what we know: Apparently it’s not enough for Nvidia’s stock to go to the moon — …

Meta is accelerating its MTIA program In sum – what we know: Meta has laid out a roadmap for its upcoming MTIA chips, with a hefty four new chips set to …

Why AI’s thermal wall is making liquid cooling mandatory The AI hardware arms race has other problems than just raw compute — like heat. GPU makers are pushing thermal design …

As data centers hit an energy bottleneck, analog chips and in-memory computing offer a low-power alternative AI training and running large models demands massive computational resources, and the GPUs doing …

Buying up old GPUs for AI might be the way to go for some smaller AI outfits Every time Nvidia drops a new flagship accelerator, the entire AI processing landscape …

New 102.4 tbps Silicon One promises efficiency gains for hyperscalers In sum – what we know: It’s easy to focus on the GPU arms race when talking about AI infrastructure, …

How can quantization turn massive models into efficient tools without ruining their accuracy? Running large language models is expensive. The biggest ones pack hundreds of billions of parameters, each stored …

Replacing copper with optical pipes could have a significant impact on the AI data bottleneck The semiconductor industry has been following the same steps for decades, revolving around shrinking the …

FPGAs may not be as powerful as GPUs, but they’re a whole lot more flexible Field-programmable gate arrays, sit in an interesting middle ground in the AI hardware landscape, somewhere …

Connecting AI chips with interconnects is arguably just as important as the chips themselves Modern AI training has moved far beyond what any single GPU can accomplish. Training large language …