Load balancers emerged in the late 1990s, when web services began outgrowing the limits of a single server. In the early days, physical appliances from F5 or Cisco divided traffic inside data-center racks. But as the cloud era arrived, the load balancer itself became "infrastructure you have to manage," and AWS ELB abstracted that away into a fully managed service.
ELB, however, isn't a single thing. There's ALB (Application LB), NLB (Network LB), GLB (Gateway LB), and the legacy CLB — each operating at a different OSI layer and solving a fundamentally different problem