Data centers are promising demand-side flexible resources that can provide frequency regulation services to power grids. While most existing studies focus on individual data centers, coordinating multiple geo-distributed data centers can significantly enhance operational flexibility and market participation. However, the inherent uncertainty in both data center workloads and regulation signals pose significant challenges to maintaining effective operations, let alone determining regulation capacity offerings. To address these challenges, this paper proposes a coordinated bidding strategy for electricity purchases and regulation capacity offerings for multiple geo-distributed data centers in electricity markets. This strategy expands the feasible region of operational decisions, including workload dispatch, server activation, and cooling behaviors. To enhance the participation of data centers in frequency regulation services under uncertainty, chance-constrained programming is adopted. This paper presents explicit models for these uncertainties involved, starting with the Poisson-distributed workloads and then addressing the unpredictable regulation signals. Numerical experiments based on real-world datasets validate the effectiveness of the proposed strategy compared with state-of-the-art strategies.