New silicon designs promise higher throughput for training workloads
Several chipmakers unveiled roadmaps for denser AI accelerators aimed at training and inference clusters.
Industry analysts said supply chain capacity remains the limiting factor, while software stacks continue to converge around open tooling. Cloud providers are expected to absorb most early volume before enterprise on-prem deployments catch up.
GE Aerospace is preparing to assemble the first XA102 adaptive cycle engine test asset for the US Air Force’s NGAP progr…
Interesting Engineering · —
MIT has designed a nanoscale computing device uses soft, flexible polymers to process information within a single micros…
Interesting Engineering · —
HiTHIUM unveils a 4MWh sodium-ion storage system with a 30-year life, targeting large-scale energy storage from 2027.
Interesting Engineering · —
China has taken a major step toward commercial humanoid robot production after UBTECH began operations at its new indust…
Interesting Engineering · —
Elisa de Martel, who left her position as chief financial officer at Alphabet's autonomous vehicle company Waymo in Janu…
TechCrunch · 57 minutes ago
I used an open-source map of a fruit fly’s brain to vibe code a website called PitchFly. Its headline suggestions were d…
WIRED · 60 minutes ago