Uploaded July 2026 | Updated September 2026, 3 hours ago
For 20+ years, companies tried to build dataflow chips — and every single one failed.
Intel tried. HPE tried. Others tried. The concept existed in academic literature. The math worked. But no one could crack the programmability problem: how do you tell a chip what to do when there are no instructions?
Elad Raz explains how NextSilicon solved what everyone else couldn't — using software algorithms that autonomously analyze your code, understand the computation, and wire the chip around it at runtime.
The idea wasn't new. The execution was.
🔔 Subscribe to keep up with the most important breakthroughs in semiconductor technology and AI compute.
Watch the full episode → bit.ly/3R4FAck
#Dataflow #ChipArchitecture #NextSilicon #Semiconductors #AIHardware #EladRaz #ComputeHistory #TechInnovation #Invested #DeepTech
For 20+ years, companies tried to build dataflow chips — and every single one failed.
Intel tried. HPE tried. Others tried. The concept existed in academic literature. The math worked. But no one could crack the programmability problem: how do you tell a chip what to do when there are no instructions?
Elad Raz explains how NextSilicon solved what everyone else couldn't — using software algorithms that autonomously analyze your code, understand the computation, and wire the chip around it at runtime.
The idea wasn't new. The execution was.
🔔 Subscribe to keep up with the most important breakthroughs in semiconductor technology and AI compute.
Watch the full episode → bit.ly/3R4FAck
#Dataflow #ChipArchitecture #NextSilicon #Semiconductors #AIHardware #EladRaz #ComputeHistory #TechInnovation #Invested #DeepTech










