From FIPS 180-4 § 5.1.1, the padding used for the SHA family of hashes begins with a binary 1, followed by a number of 0s, and finally a 64-bit representation of the message length:
Suppose that the length of the message, M, is 𝓁 bits. Append the bit "1" to the end of the message, followed by k zero bits, where k is the smallest, non- negative solution to the equation 𝓁 + 1 + k ≡ 448 mod 512. Then append the 64-bit block that is equal to the number 𝓁 expressed using a binary representation.
A form of this question has already been asked before, and the answer seems to be that it's likely because MD5 did that, and MD4 before that. But while that explains why MD5, SHA-1, and SHA-2 do it, it leaves the initial question unanswered. What is the purpose of starting the padding with a 1? If the correct answer for the SHS is "because that's how MD4 did it", then why did MD4 do it?
For SHA-3 which does not use length encoding, the purpose is to eliminate ambiguity:
The initial 1 bit is required so messages differing only in a few additional 0 bits at the end do not produce the same hash.
MD4, which has always employed length encoding, still seems to use this padding.