This research provides empirical evidence on whether decentralization offered by blockchain technology can mitigate the problem of “too connected to fail” in crypto ecosystem by comparing centralized and decentralized exchanges. The Minimum Spanning Tree measures changes in network topologies of crypto token pairs across crypto exchanges during normal and crisis periods, and Conditional Value at Risk (ΔCoVaR) is employed to investigate their systemic risk contribution. Finally, regression analysis is performed to ascertain the link between centrality values and systemic risk contribution. Results showed that centrality values significantly impact systemic risk contribution of token pairs listed on centralized exchanges. Conversely, insignificant results were found for all token pairs traded on decentralized exchanges. This confirms that decentralized mechanism mitigates systemic risk propagation arising from returns interconnectedness among crypto assets and institutions. However, other factors such as liquidity, smart contract vulnerabilities, and regulatory compliance might affect risk characteristics of decentralized assets and platforms.