According to the CDC there have been 12,000 deaths since 1990 from vaccines, from their numbers it was averaging just over 150 a year, this past year was/is 6,887 but I would expect that because of the number of vaccines given.
Just the annual flu vaccine in the U.S. reaches some 200 million doses. And there's a lot of childhood and young adult vaccines administered, too. If we vaccinate everyone in the U.S. with a two-part vaccine like Pfizer, that would be ~700 million. So if Covid vaccines were remotely normal, a jump of 3x or 4x or so in VAERS would be reasonable. Not a jump of 40x.
Yes, VAERS data quality is not great, but it exists because it has been useful in the past. Where there's smoke, there's often fire.