2020 contains two separate stories that will define the rest of the decade.
COVID-19 forces the largest experiment in digital transformation ever run. Companies that had been deferring cloud migration and digital commerce for years are compelled to execute in weeks. Supply chains shift, physical retail contracts and ecommerce volumes surge in ways that expose which platforms were genuinely scalable. For enterprise software teams, the year tests whether the systems they built can handle load, configuration changes and new integration requirements under pressure.
GPT-3 is released in June 2020, and nothing in AI is the same afterward. With 175 billion parameters — more than two orders of magnitude larger than GPT-2 — the model demonstrates few-shot learning: the ability to perform new tasks from a handful of examples in the prompt, without retraining. Developers with API access build working prototypes of text completion, code generation and question answering in hours. The practical implication — that a single large model can perform diverse tasks without task-specific training — reshapes thinking about how AI gets built and deployed in products.
2020 establishes both the infrastructure baseline (remote-first, cloud-native, resilient) and the AI capability baseline (large language models as general-purpose tools) that the rest of the decade will build on.