Anker Innovations Unveils World First THUS AI Chip Platform and a Multi-Year Plan to Bring Local AI to Consumer Devices
Anker Innovations has unveiled world-firsts
THUS, a breakthrough AI chip platform that fundamentally changes how artificial
intelligence runs inside consumer devices.
Rather than sending data back and forth
between memory and processor, the way virtually every modern computing system
has operated for decades, THUS™ performs AI computation directly where the data
already lives. The result is dramatically lower power consumption, faster AI
processing and the ability to run significantly larger neural networks inside
ultra-compact devices previously considered too constrained for advanced AI.
The breakthrough marks the world’s first
Neural-Net Compute-in-Memory (CIM) AI audio chip, debuting inside earbuds, one
of the most technically demanding environments in consumer electronics.
“Every AI chip built until now stores
the model on one side and does the computation on the other,” said Steven Yang,
Founder and CEO of Anker Innovations “To think, the device has to carry all
those parameters across, many times per second, every single inference. THUS™
puts the computation where the model already lives. The model never has to move
again.”
For decades, modern computing has relied
on the John von Neumann architecture, where memory and processing exist
separately. While effective for traditional computing, the model becomes
increasingly inefficient for AI workloads, which require constant access to
enormous volumes of neural network parameters. That movement of data consumes
energy, creates latency and limits the size of AI models that can operate
inside battery-powered consumer devices. THUS™ solves this problem using a
compute-in-memory architecture inspired by the human brain, where storage and
computation happen simultaneously in the same place.
By embedding AI computation directly
inside NOR Flash memory cells, THUS™ eliminates the need for constant data
transfer between memory and processor, allowing AI models to remain stationary
while computation occurs locally.
Leave A Comment