In the years since engineers first began making circuits using transistors (three-terminal devices made of a semi-conducting materials like silicon) the number of transistors on a circuit board has followed what is called Moore’s Law.

Gordon E. Moore, co-founder of Intel Corporation, observed in 1965 that the number of transistors on an integrated circuit chip doubled roughly every two years, and he predicted that this trend would continue for at least ten years. Surprisingly, this trend has continued all the way to the present. Complementary metal-oxide-semiconductor (CMOS) transistors are most commonly used for computing. These transistors are actually a pair, made of one n-type transistor, through which current flows via an excess of electrons, and one p-type transistor, through which current flows due to “holes,” or a lack of electrons.

Today’s CMOS transistors have feature sizes, or the distance between terminals, of less than 30 nm, which is only a handful of silicon atoms. According to EE Times, this is the point at which Moore’s law finally begins to break down, as speed and efficiency of smaller transistors has ceased to reduce cost.

So what is the next step in the future of computing? IBM has designed a computer chip called TrueNorth that attempts to mimic the neural network of biological brains. By using digitally programmable neurons and synapses, TrueNorth combines memory, computation, and communication into a single package. Unlike current computer chips, these various components are already interconnected, so a single bridge line, which can only access instructions, data, and operational information one at a time, is no longer needed. By removing this operational bottleneck, speed and efficiency can be increased.

This new neural chip still uses CMOS technology, but in a new way, which is both faster and more efficient in power consumption. The computer clock, the traditional trigger that coordinates computing processes, has been replaced by what is called event-driven computing. That is, the neurons react to signal spikes asynchronously rather than to regimented clock cycles. These signal spikes may be output by either neural or digital signals, allowing the chip to react to external stimuli as needed and removing the need for a regimented, cyclic reaction. Finally, to reduce size, TrueNorth is entirely digital.

Unlike the human brain and conventional computer chips that try to mimic brain technology, TrueNorth was designed without analog circuitry, allowing it to use the 28nm transistors designed for cell phone use.

TrueNorth has been the product of funding from the US Defense Advanced Research Projects Agency (DARPA) since 2008, and is the result of over a decade of research on the part of Dharmendra Modha, the project lead. TrueNorth was designed as a commercial product, so corners were cut when it came to trying to gain an understanding of the biological brain. Other research projects, such as SpiNNaker, are looking at neural computing as a way to understand the human brain and how its complex processing can be mimicked to create technology that can “learn” and adapt to its surroundings.

Ultimately, the research team at TrueNorth and other neural computing companies hope this technology will be commercially viable, and that TrueNorth and similar neural chips will soon be powering cell phones, smart cars and anything else requiring a computer. According to IEEE Spectrum, IBM and others hope to create a faster, better technology and continue to change the way computers affect our everyday lives by using less power to do more, and by creating a brainlike technology that can “learn.”