Mar. 26 at 10:25 AM
$GOOG Global sell off due to Alphabets TurboQuant. What the general public doesnt know about is Neuromorpic computing but they will imo.
Neuromorphic computing, conversely, does not try to compress data to fit through a bottleneck like TurboQuant; it eliminates the bottleneck entirely. By mimicking the human brain, neuromorphic chips (like
$INTC Intel’s Loihi 2,
$IBM IBM’s NorthPole or
$BRCHF Brainchip's Akida) integrate memory and processing into the same neurons and synapses.
If TurboQuant makes AI significantly cheaper to run on standard hardware, it will likely accelerate the software side of AI (ie neuromorphic spiking neural networks, etc.).
As AI becomes more efficient, the demand for edge AI (AI on sensors, watches, and remote devices) will explode. This is the primary domain of neuromorphic computing, which excels at low power edge tasks.
Advances like TurboQuant prove that sparse and compressed data models work. This validates the neuromorphic approach even more imo