IBM and its key research partners managed to create a new type of a transistor that could theoretically fit 30 billion times into a 5-nanometer chip, which it also developed a manufacturing process for. Essentially, all that's left is to actually use the tiny transistors with the new process to produce a proof-of-concept 5nm chip upon which to base commercial production models. The new advancement involved help from GLOBALFOUNDRIES and Samsung, and could enable enhanced computation abilities over current devices, with energy savings to produce two or three times the total runtime, depending on the device in question, IBM claims. The company will be showing off the new process in detail at the upcoming 2017 Symposia on VLSI Technology and Circuits that started today in Kyoto, Japan.
A little less than two years ago, 7nm chips hit the scene, and while major chipmakers like Samsung and SK Hynix are currently in an arms race with those, 10nm chips are the most popular option on the market right now. 5nm chips, especially those with 30 million transistors as planned by IBM, stand to deliver far more power in a smaller, more design-friendly package, with less heat and energy consumption. Some of the largest gains will likely be seen in fields that require a small or power-efficient package, including mobile and Internet of Things (IoT) devices. The 5nm breakthrough process takes advantage of nanosheets, a technology that vastly reduces the base size of a chip, allowing for smaller transistors and more pathways in a smaller footprint.
For the time being, 5nm chips are merely a concept design with the technology to back that design up. Though IBM and its partners will likely produce test chips and perhaps ship their own commercial products based on the model in the future, the real gains will come from other players in the semiconductor industry licensing the design out to create 5nm chips for commercial applications, and even putting their own spin on the design. It's worth noting that certain difficulties regarding the physics of light and electricity are still in the way of the creation and eventual commercialization of a 5nm design, and things are unlikely to get any tinier than this without a significant breakthrough of some sort in the future.