Quantum computing promises significant enhancement in computational capabilities, yet it is hindered by noise and limited qubit counts, restricting us to Noisy Intermediate-Scale Quantum (NISQ) devices. While progress is being made to achieve quantum advantage with NISQ through approaches like variational algorithms and error mitigation, understanding their fundamental limitations remains a critical challenge. In my talk, I will explore the computational boundaries and entanglement capacities of NISQ devices without error correction. Our research indicates that once circuit depths surpass $\omega(\log(n))$, NISQ devices fail to outperform classical counterparts as determined by polynomial-time algorithms, ruling out the use of prominent quantum algorithms and simulations in this context. Additionally, we identify a logarithmic upper limit on entanglement within one-dimensional qubit arrays, highlighting the entanglement scalability constraints of current quantum systems. This talk is primarily based on the work [arXiv:2306.02836].