Under the concept of atomic settlement, the time between the trade and the settlement of transactions is reduced to a very small fraction of time (e.g., seconds or less), making them instantaneous. A second feature is the exchange of all the assets in the transaction in a single event, making them simultaneous — no delivery or payment occurs if the others do not occur too, commonly known as delivery versus payment (DvP) or payment versus payment (PvP). Making the settlement of the transaction instantaneous and simultaneous may have several benefits. It could effectively mitigate risks to all parties by guaranteeing the immediate and concurrent transfer of all assets in a transaction. By squeezing the transaction to its lowest time and space dimensions (i.e., its atom), some risks related to the settlement of transactions may be mitigated, e.g., replacement-cost risk. However, making the settlement instantaneous and simultaneous only benefits some trading scenarios. In other scenarios, it will create higher liquidity requirements for the parties involved and increase settlement risk. This could be particularly burdensome in a positive interest rate environment. Therefore, notwithstanding its current allure, it is important to identify in which cases it makes no sense to settle atomically and how to make it work. For those transactions that do not benefit from instantaneous and simultaneous settlement, we suggest not working on the smallest units (i.e., the atoms) but on the second smallest units (i.e., the molecules). We propose building an intermediate centralized layer between the trade and settlement layers that will use atoms (i.e. single transactions) to use smart algorithms to create molecules (i.e., sets of transactions) that make sense to settle from a liquidity-efficiency viewpoint. This way, more transactions will benefit from faster and concurrent transactions while avoiding high liquidity costs.