Deep Mind's latest protein-folding model is impressive, yet accessibility is still constrained by cost of compute. Distributed alternatives can deliver comparative models more efficiently.