Samsung has announced that its chip factory in Texas will play an integral role in the quest to get 4nm processors to market by 2020, and has invested $1 billion into that factory to help make it happen. The Texas foundry boasts production of advanced chipsets for devices like smartphones and tablets on a grand scale, making it one of the best research and development hives that Samsung has available. Since that particular factory also specializes in NAND memory and ultra-thin LIS chipsets, it is safe to assume that Samsung will be expanding its presence in these markets, as well. The Austin, Texas facility has received roughly $16 billion in funding from Samsung since it opened in 1997, and managed to put some $3.6 billion back into the local economy in 2015 alone.
Samsung is already hard at work designing and implementing processes for a 7nm mobile chipset, though rumors point to production on that tech not finishing in time for it to be used in the company's Galaxy S9 flagship handset, which is likely to come out next spring. For the time being, Samsung is still producing and marketing 10nm chips, including the Exynos 8895 that sits at the heart of the recently released Galaxy S8 outside of the United States where Samsung uses Qualcomm's Snapdragon 835 processor. Samsung's roadmap indicates the development of an 8nm process this year, a 7nm process in 2018, and a 4nm one in 2019, meant to be ready for use in 2020. If the goal is indeed to produce a 4nm chip by 2020, that would mean that Samsung has about two years to cut die size in half. This implies a major breakthrough is on the horizon, which is actually the case in the industry right now.
There are significant barriers to getting chipset processes down to 5nm and under, mostly pertaining to the quantum and electrical physics involved in the impulse transit that makes processors move and execute instructions. Those barriers were significantly weakened recently when IBM created a theoretical manufacturing process for 5nm and prototypes of an extremely small transistor to be used in that process. Researchers still have not quite cracked 5nm, but IBM's achievement is a step in that direction. Once that barrier is fully broken, getting processors down to 4nm will be the next step.