Memory demands of RunPCA on fairly large dataset

[lizchcase]

I am analyzing a large dataset of >350,000 nuclei using Seurat (rds file is ~886,000 kb, not sure how large it is when read into Seurat). I ran the following code:
data <- NormalizeData(pbmc)
data <- FindVariableFeatures(pbmc, selection.method = "vst", nfeatures = 2000)
data <- ScaleData(data)
data <- RunPCA(data , verbose = F, npcs = 50)
data <- RunUMAP(data , dims = 1:50, verbose = F)

However, on the RunPCA step, I got stopped because there was not enough memory on the server ([1] "Running PCA..." / [1] "Calculating PC distance matrix..." / Error: cannot allocate vector of size 1657.4 Gb)).

My questions are:

1. would changing NormalizeData, FindVariableFeatures, and ScaleData to SCTransform reduce the computational needs here?
2. would you expect that a dataset of this size would require this much memory for RunPCA?

Thank you!