Yes two way communication is not possible with a spoofed IP, and so hackers usually try other ways of masking their IP, like using VPNs and Tor
However, some attacks do not require any reply from the server so they are possible to carry out with a spoofed IP. The simplest example of this would be a DDoS, where the bots in a botnet could spoof their IP to bypass IP blacklisting (and perhaps also conceal their origin).
Generally, vulnerabilities that can be exploited over UDP would not require any reply from the server so IP spoofing may be used. For example, CVE-2002-0649, which the SQL Slammer worm employed, could be exploited with a single UDP packet and no reply was required from the server. Theoretically, an attacker could use such a vulnerability with a spoofed IP to achieve remote code execution. Of course, this assumes that the attacker already knows if and where such a vulnerability exists in the target.