AMPK is a promising target for various chronic illnesses such as diabetes and Alzheimer's disease (AD). We sought to develop a novel small molecule that directly activates AMPK, with the potential to fundamentally modulate the pathogenic mechanisms of the metabolic disorders. To identify a potent novel pharmacophore in an unbiased way, we performed structure-based virtual screening on a commercially available chemical library, and evaluated the actual AMPK activity of 118 compounds selected from 100,000 compounds based on docking scores. Additional iterative molecular docking studies and experimental evaluation of AMPK activity led us to select a hit compound, B1, with a chromone backbone. Using the hit compound and other compounds structurally similar to the hit compound, we identified the chalcone structure as a new scaffold with more efficient interactions with key residues required for AMPK activation. From the newly designed and synthesized chalcone derivatives, we discovered compound 6 as a candidate compound. Compound 6 showed the most efficient interactions with the key residues of AMPK at in silico study and demonstrated significant activation of AMPK in both in vitro and in cellular assays.