Many molecular design tasks within computational protein design and computer-aided drug design can be reduced to free energy optimization problems. Alchemical free energy methods provide high accuracy free energy predictions from molecular dynamics simulations due to their rigorous basis in statistical mechanics. Consequently, alchemical methods have been widely adopted by the pharmaceutical industry, but are relatively unexplored for protein design. Traditional alchemical methods are too inefficient for protein design because they can only compare pairs of sequences, however an emerging alchemical method known as λ dynamics allows exploration of high dimensional sequence spaces within a single simulation. Thus λ dynamics possesses the scalability required for protein design. Retrospective benchmark studies of T4 lysozyme demonstrate the accuracy of λ dynamics for protein design. Prospective studies in ribonuclease H showcase the ability to predictively explore high dimensional sequence spaces. Methodological developments underpinning these benchmarking studies will also be presented, along with parallel benchmarks for computer-aided drug design of small molecules. The talk will conclude with recent developments enabling exploration of higher dimensional chemical spaces and the ongoing projects they are driving in site saturation mutagenesis, high throughput screening, and concurrent cofactor and sequence design of redox enzymes.