Your best bet is likely to combine your datasets.
First we'll combine the data into one using the tidyverse and do a bit of reshaping:
library(tidyverse)
groupA = tibble(inst1 = 50, inst2 = 20)
groupB = tibble(inst1 = 80, inst2 = 60)
all = bind_rows(
mutate(groupA, group = "A"),
mutate(groupB, group = "B")
) %>%
pivot_longer(inst1:inst2)
Now we can plot:
ggplot(all, aes(x = value, y = name, fill = group)) +
geom_col() +
labs(y = NULL) +
theme(legend.position = "top")

or...
ggplot(all, aes(x = value, y = name, fill = group)) +
geom_col(position = position_dodge()) +
labs(y = NULL) +
theme(legend.position = "top")
