Eventually, the Bush White House admitted it had lost 22 million emails, not 5 million. Then, in December 2009—well into Barack Obama’s administration—the White House said it found 22 million emails, dated between 2003 and 2005, that it claimed had been mislabeled. That cache was given to the National Archives, and it and other plaintiffs agreed, on December 14, 2009, to settle their lawsuit. But the emails have not yet been made available to the public.
The Washington Post, in an editorial, accepted the White House explanation that the emails could have been lost due to flawed IT systems.
IIRC, archiving executive emails is required, but access to them is restricted until years later (2020 or thereabouts for GWB).. the thinking being that it is in the national interest to document the ongoings in the white house and allowing earlier access would lead to abandoning email and loss of the discussions/decision-process.