AI coding will be more expensive than human developers
A forecast predicts that by 2028, using AI for coding tasks will become more expensive than employing human developers, citing rising costs of AI inference and compute resources as key factors.
Background
- The article discusses a forecast that by 2028, using AI to generate code may cost more than employing human software developers, due to the high computational and energy costs of running large language models (LLMs) at scale.
- "AI coding" refers to tools like GitHub Copilot, Cursor, or ChatGPT that write or suggest code based on natural language prompts. These tools rely on massive AI models that require expensive hardware (GPUs) and electricity.
- The prediction comes from an analysis of the total cost of ownership (TCO) for AI-assisted development, factoring in inference costs, model training amortization, and infrastructure — not just current subscription prices.
- This challenges the current narrative that AI coding is cheap or free; the hidden costs (server farms, chips, cooling) are borne by the provider now but may be passed to customers as adoption grows and subsidies end.
- Key context: many tech companies currently offer AI coding tools at below-cost prices to gain market share and improve models. The forecast suggests this is unsustainable.