Since there's a lot of misinformation in pretty much all articles about this, here's some context:
Jingjia micro developed the very first domestically manufactured GPU, the JM5400 (GT 210 40nm), in 2014 based on the 65nm node in order to replace ATI 780, M54, M72, M96, etc. cards used in military equipment and aircraft. After the success of JM5400, they managed to R&D the 28nm node. In 2017, they announced the JM7200 (GT 640 28nm), based on the 28nm, which the company planned to release to the consumer desktop market. It released in 2018 to a major success, and they went to the drawing board to refine the node even more to help them deliver GTX 1080 performance level domestically. Of course, this won't come cheap as the company sought investment for the R&D process (880M Yuan, or around $150 million dollars at the time, which is quite cheap compared to, as the company claims, the several billion dollars nVidia invested in the R&D process to bring Pascal architecture--and its cards' performance--to the world), and they already secured the funds to start the process. They hope (as they postponed the release of JM7200 by 2 years) to release a GTX 1050-like card by 2020, and a GTX 1080-like card by 2021 (including every other card between the two).
TLDR; a Chinese company wants to disrput the domestic GPU market in China and replace foreign companies (AMD and nVidia). In only 3 years (starting from basically zero) they managed to go from GT 210-performance on a 65nm node to GT 640-performance on a 28nm. They want to reach GTX 1080-performance on a refined 28nm node by 2021.
Jingjia micro developed the very first domestically manufactured GPU, the JM5400 (GT 210 40nm), in 2014 based on the 65nm node in order to replace ATI 780, M54, M72, M96, etc. cards used in military equipment and aircraft. After the success of JM5400, they managed to R&D the 28nm node. In 2017, they announced the JM7200 (GT 640 28nm), based on the 28nm, which the company planned to release to the consumer desktop market. It released in 2018 to a major success, and they went to the drawing board to refine the node even more to help them deliver GTX 1080 performance level domestically. Of course, this won't come cheap as the company sought investment for the R&D process (880M Yuan, or around $150 million dollars at the time, which is quite cheap compared to, as the company claims, the several billion dollars nVidia invested in the R&D process to bring Pascal architecture--and its cards' performance--to the world), and they already secured the funds to start the process. They hope (as they postponed the release of JM7200 by 2 years) to release a GTX 1050-like card by 2020, and a GTX 1080-like card by 2021 (including every other card between the two).
TLDR; a Chinese company wants to disrput the domestic GPU market in China and replace foreign companies (AMD and nVidia). In only 3 years (starting from basically zero) they managed to go from GT 210-performance on a 65nm node to GT 640-performance on a 28nm. They want to reach GTX 1080-performance on a refined 28nm node by 2021.