Is the Global Neuromorphic Chip Market Hindered by Moore’s Law?


Published Date : Feb 29, 2016

Neuromorphic chips are supposed to be the perfect integration of VLSI systems with analog designs to create what science fiction movies have been showing for a long, long time. The birth of artificial intelligence was inevitable, as humankind’s drive to develop smarter technologies is on a rapid rate of increase. Ideally, a neuromorphic chip is a copy of the human brain. These are hardware chips that contain oxide-based memristors, transistors, and threshold switches, which are together capable of showing a real time learning curve when exposed to knowledge and external data. We can, therefore, imagine that the demand for the global neuromorphic chip market is high and will continue to increase.

Recent Developments in the Global Neuromorphic Chip Market

Darwin: Spiking neural networks are supposed to be an advanced aspect of the global neuromorphic chip market and artificial neural networks in general. Biologically speaking, they are more realistic than classic neuromorphic chips and can even perform better than them at lower power consumption rates. A group of Chinese researchers has recently created the neural processing unit called Darwin. It is a neuromorphic hardware co-processor which uses SNN and is created with the help of CMOS tech.
DARPA Synapse Program: In 2008, the DARPA Synapse program helped create the True North chip with IBM. This was a device which mimicked 256 programmable neuron-like architectures and are very much similar to the human brain’s synapses.
Use of New Materials: Researchers, including those at IBM, are considering the implementation of silicon photonics and carbon nanotubes. These materials could theoretically give the neuromorphic chips a much needed boost in speed without needing the use of advanced quantum mechanics.

Traditional computing systems employ a processor, a storage system, and a memory system, with all three connected by a bus. This conventional architecture is showing signs of bottlenecking, as the growth of the amount of data that needs to be handled is overtaking the current rate of development in conventional chip designs and hardware architecture. Industries are therefore looking to the global neuromorphic chip market for answers which can help resolve the critical issue of data management in a smart, efficient, and accurate manner.