According to a Gartner forecast, by 2028, the costs of using generative AI for coding could exceed the average global developer salary, which is approximately $2,000 per month.
What Happened
Gartner predicts an economic turning point in software development: the costs of using LLMs for coding may become higher than the costs of hiring human specialists. The main drivers of rising prices are increasing token consumption, the transition to consumption-based billing models, and high infrastructure costs for computation (inference/compute).
Context
The current transition from using AI as an auxiliary tool to a full-fledged managed infrastructure (model-as-a-service) carries the risk of uncontrolled growth in operating expenses (OpEx). Companies like Uber are already facing the problem of rapidly depleting token budgets in the absence of clear return on investment (ROI).
Why It Matters for the Industry
Industries need to move from uncontrolled AI implementation to regulated operating models. This implies the development of AI Observability class tools and LLM Gateways to manage quotas, prompt caching, and automatic task segmentation between cheap (small) and expensive (large) models. In the long term, an AI FinOps market will emerge, where managing token costs becomes as critical an engineering discipline as managing latency or memory.
Why It Matters for Users
Developers and technical managers should focus on prompt optimization and careful model selection to balance quality and cost. It is important to implement strict monitoring of API usage and token limits into current workflows to prevent the cost of AI assistants from becoming an uncontrolled budget item.
What Is Not Yet Known / Limitations
There is a difference in risk assessment focus: engineers are focused on operating expenses (OpEx), while product creators see the need for new infrastructure management tools.
Sources
Author
Look at AI, Editorial Team