Enterprise AI Spending Surges Despite Plummeting AI Unit Costs

Gist While the per-unit cost of running Artificial Intelligence AI has plummeted exponentially, enterprise spending on AI is surging at an unprecedented rate. This paradox is driven by a shift in usage from simple, single-prompt chatbots to complex "AI agents" that perform multi-step, reasoning-heavy tasks requiring vastly more computing power. For civil services aspirants, understanding this dynamic is crucial, as the economic viability of integrating AI into public governance and private enterprise will depend on strategically mixing open-source and proprietary models to manage skyrocketing infrastructure costs. Background To understand the economics of AI, one must understand how AI usage is billed. The foundational unit of AI processing is a "token" pieces of words or data . Historically, firms relied heavily on Proprietary Closed Models like OpenAI’s GPT or Google’s Gemini, where the developer controls the parameters and charges users via an API based on token consumption. Recently, Open-weight Models like Meta’s Llama have disrupted this space. These allow organizations to download and run the trained parameters themselves, though they are not fully "open-source" as the original training data may remain…