The rapid expansion of cloud services and the increasing reliance on them have made load balancing a significant research challenge. As technological services grow in demand, optimizing their performance becomes essential. This study addresses the load-balancing challenge by mathematically formulating the problem within cloud computing systems, where an objective function is developed to minimize response time, computational cost, and load imbalance, while ensuring constraints such as task allocation to individual virtual machines (VMs) and adherence to resource limits. To solve this problem, we propose a novel hybrid algorithm, ACOCSA, which combines ant colony optimization (ACO) and crow search algorithm (CSA). Our experimental results indicate substantial performance improvements. Specifically, ACOCSA reduces response time by 12% compared to ACO, achieving a reduction from 8.5 to 7.48 s for 400 tasks. Additionally, it demonstrates a 5% improvement over GIJA, with response times of 8.20 s compared to 8.62 s. ACOCSA also completes tasks 33.3% faster than ACO, reducing task completion time from 450 to 300 s for 100 tasks, and 7.8% faster than GIJA, which requires 324 s. The average cost is reduced by 12.5% when compared to CSA, with a cost of 0.11 versus 0.125 for 10 tasks. Furthermore, ACOCSA achieves a 3.5-point increase in fairness index, from 86 to 89.5, across 100 iterations, indicating improved load distribution and balanced VM utilization. These findings demonstrate that ACOCSA outperforms existing algorithms in terms of response time, cost, and fairness of load distribution. Statistical analyses confirm that ACOCSA consistently achieves superior load balancing efficiency, ranking first among other methods, with a top mean rank of 1.00 in the Friedman test (p value < 0.000047). Although further empirical validation is needed to explore its energy-saving potential, the results emphasize ACOCSA’s suitability for real-world dynamic cloud environments.