Massive MIMO produces a 2-7X capacity improvement, far more than the 15-50% of 5G NR software. Every engineer knows the 5G hype has become ridiculous, especially because Massive MIMO is a key component of 4G.
Guido Weissbrich of Vodafone Germany confirms they have moved from trials to deployment. His colleagues are in production in Australia. Bharti in India is covering much of Kolkata, formerly known as Calcutta. Verizon is widely deployed in the U.S. but has released few details.
All of these, I believe, are below 2.1 GHz and are FDD. One vendor off the record guessed FDD Massive would only be a 1.5X improvement. Karri Kuoppamaki of T-Mobile expects a 2-3X improvement.
FDD performance remains unproven and I'd welcome data. Marzetta tells me the solution is to change everything FDD to TDD, but that's not easily done.
Sprint in the U.S. is going hard in TDD 2.5 GHz, including New York. They are reporting an average 4X improvement. That's less than Softbank Japan (up to 10X) and more than China Mobile (3X.) But performance varies so much by terrain comparisons are misleading. Below millimeter wave, Massive MIMO is improving capacity far more than "New Radio," which the marketing and lobbyist use to define 5G.
Extensively rewritten Aug 8 for clarity.