In fifth generation and beyond (B5G) technologies, delay constrains emerge as a topic of particular interest for ultra reliable low latency communications (e.g., enhanced reality, haptic communications). In this short paper, we study the performance of a two-user uplink NOMA network under quality of service (QoS) delay constraints, captured through each user's effective capacity (EC). We propose novel closed-form expressions for the EC of the NOMA users, validated through Monte Carlo simulations. Interestingly, our study shows that in the high signal to noise ratio (SNR) region, the "strong" NOMA user is more penalized in terms of its EC under the same delay constraint as the "weak" user.