On-Demand Instances are recommended for short-term, irregular workloads that cannot be interrupted. The company needs capacity for only 24 hours once per year, so it would not be economical to make a one-year or three-year Reserved Instance commitment for such limited usage. With On-Demand pricing, the company pays only while the instances are running, without a long-term commitment. Therefore, option A is correct. Spot Instances can provide significant discounts but use spare EC2 capacity that AWS can interrupt. Because the workload explicitly cannot be interrupted, Spot Instances do not meet the requirement. Reserved Instances can reduce costs for steady, predictable, long-running usage, but paying for a commitment to cover an annual 24-hour workload would generally be less economical than paying the On-Demand rate for that brief period. Dedicated Instances run on hardware dedicated to one customer and normally cost more; the question includes no compliance, tenancy, or licensing requirement that would justify dedicated hardware. The decisive combination is “cannot be interrupted” plus “short-term and infrequent,” which maps directly to the documented On-Demand use case.