The page explains this with a small note: "Race/Ethnicity data were available for 64% receiving at least one dose and 69.1% of people fully vaccinated". So increasing the numbers by about 50% to account for that makes it look more reasonable.
Regardless, if we're nitpicking data quality surely a random nbc news poll doesn't exactly pass muster either.
The NBC News poll is designed to be representative of national numbers and the CDC page specifically calls out that it isn't designed to be representative of national numbers. Therefore the CDC numbers aren't necessarily more accurate just because the sample size is bigger.
And like I said, even if we take the CDC numbers as the objective truth, they do not contradict my original point.
Regardless, if we're nitpicking data quality surely a random nbc news poll doesn't exactly pass muster either.