Neighborhood at atomic scale is important for the properties of advanced alloys. The preference or avoidance between neighboring atomic species is known as chemical short-range order (SRO). While SRO in metallic medium/high entropy alloys has garnered substantial attention recently, understanding SRO in semiconductor alloys remains underdeveloped. Motivated by theoretically predicted SRO and its dramatic impact on band structure, here we perform statistical analyses of atom probe tomography data to quantify SRO in SixGe1-x-ySny alloys (x<12 at.%, y<16 at.%). Leveraging a side-by-side experiment-theory comparison at the same spatial scale enabled by machine-learning neuroevolution potentials of first-principles accuracy, we reveal, for the first time, a notable SRO favoring Si-Si 1st nearest neighbors even at dilute Si and Sn compositions. The SRO can be tuned by varying precursors, offering a new degree of freedom for band engineering beyond composition and strain, and enabling new phase-change materials based on SRO transitions for Si electronics/photonics.