This sounds like typical flawed math used to hype these academic studies.
They're comparing everything based on the power consumption of the in-use phone, i.e. manufacturing is x% of total impact. We know phones use a tiny amount of power and it's better for phones to use less power, both environmentally and for convenience. Therefore, this number will only go up.
Say I had a phone that manufacturing and end of life was 75% of total impact and in-use emissions was 25%. I change nothing, but I magically make it consume no wall power at all. Now manufacturing and EOL is 100% of total impact. This article would say that's worse. However, my net impact went down 25%, so it's obviously better.
And how does this impact compare to say, people who drive to the Apple store 10 miles to buy the phone? Why aren't they giving the answers in grams CO2e?
They state ICT energy consumption is going up. What is it replacing? Online billing replaces paper bills. Video conference replaces travel. Downloads replace sending plastic disks. Are net emissions going up or down? We know that in the US, absolute demand for electric power is flat or slightly decreasing, and per capita power consumption is decreasing.
In other words, fake news.