The soil does not need to be millions of meters deep in order for the forest to be both a net carbon sink and net positive oxygen producer -- it can slowly acrue over time.
And even if biological respiration processes in the forest began to equal carbon sequestration processes -- the forest would still be a massive carbon sink over its lifetime given the massive quantites of plant matter and soil present.
Yes, the rainforest has been a massive carbon sink over its 30+ million year lifetime - because of the carbon it sequestered while it was first growing and expanding.
Yes, it could be producing very, very tiny amounts amounts of oxygen each year so that the rise in soil level isn't noticeable. If the rainforest was still slowly sequestering carbon in soil at a rate of about 1 meter of soil depth per 30 million years, then you're talking about it sequestering about 0.00005% of the carbon it captures through photosynthesis. If you're calling the article "completely wrong" over that, I'm not sure what to say.
And it's not just the yearly rate of oxygen production that must be negligible. The total oxygen produced over the Amazon's lifetime is negligible. As the article points out, if you burned the entire Amazon rainforest, releasing all the sequestered carbon, it would only decrease atmospheric oxygen from about 20.95% to 20.93%.
The biomass of the Amazon is not, for the most part, sinking into very salty water.
If 3kg of carbon sequestration per year per square meter were accumulating as soil, and the rainforest has existed at least 30 million years, the soil would be over 2,000,000 meters deep (with rich organic soil being about 9% carbon). If even 1% of the carbon captured through photosynthesis was sequestered in soil, we should still have much, much deeper soil.
