Stanford Professor Paulraj told me in 2014 that MIMO would likely increase wireless capacity 50X. Other world-class engineers discussed 100X improvements.
ZTE's 128-antenna Massive MIMO array deployed in Guangzhou delivered 10.3 Gbps down. The spectral efficiency is 100.3 bit/s/Hz, ~20X what LTE could do in 2014. It's about twice the throughput of today's typical Massive MIMO at 3.5 GHz.
Higher frequencies can use smaller antennas. The 128 4.9 GHz antennas are about the size of 4-8 antennas for 700 MHz. Broadcom's Henry Samueli was working on mmWave antennas small enough to fit on a chip. (I don't believe Henry or anyone else developed that as a commercial product.)
The reach of 4.9 GHz and higher frequencies is much less than what's commonly used today. The speeds quoted here are presumably reached very close to the antennas. Few customers will receive speeds that high.
5G standards allow up to 400 MHz, plenty of room for improvement.