No, he/she is saying that "average" is a poor metric to represent the totality of the samples. This is particularly true when you have such a high standard deviation.

By eliminating some of the biggest outliers, you're reducing the standard deviation, thus making the average metric more representative.

You'd also have to disregard the companies with very low or no value, but it seems this was already done in PG's original statement: "Of the 285 _that have_ valuations..."

Well I think more precisely, he's implying if you take out the few heavy hitters, then the value drops precipitously.