The computer industry has been facing an existential problem for a number of years now: because of the physical and quantum limitations of silicon dioxide, the speed of a silicon-based computer processor can only pushed so far before it melts. To get around this, the industry has resorted to packing more processors into individual devices, hoping that the additional processing power will make up for the lack of an increase in speeds, but this method can also only be pushed so far. Researchers are working on new technologies in the hopes of replacing silicon altogether, with optical and quantum-based processors being developed. And conversely, at least one tech startup is using a very old technology in their bid to push beyond the silicon wall.
Koniku, a biotech startup company founded by Oshiorenoya Agabi, is developing their processors based on a technology that already produces the most powerful computers on the planet — the biology of our own brains. Agabi’s ambition to revolutionize the computer industry is built on his experience in the biotech industry: “Essentially, for the last fifteen years, I have worked to understand how neurons talk to each other. I’ve worked on how to communicate with individual neurons—how to read information from them and write information into them.”
The 100 billion neurons that make up the average human brain represents more than 40 petaflops of processing power — roughly 500,000 times the power of a typical iPad — and is far more energy efficient than our current technology: The world’s most powerful supercomputer, China’s Tianhe-2, has nearly the processing power of the human brain, but uses 24 megawatts of power, as compared to the mere 10 watts that the human brain requires.
Koniku’s current prototype, dubbed the Koniku Core, is a networked grid of 64 artificial neuron shells, used to support the individual biological neurons that power the chip. The current device is capable of basic computation, and has been used to demonstrate control over a chemical-sensing drone. Agabi says that they expect to be able to miniaturize the tablet-sized device down to the size of a nickel by 2018.
While the 64-neuron prototype is capable of simple functions, he expects that a 500-neuron device could control driverless cars; 10,000 could process real-time imaging in a manner similar to our own eyes; 100,000 neurons could control robotics with integrates senses; and 1 million neurons could produce outright cognition — a machine that could think for itself.