Can tech companies learn to love cheaper AI models?
Rising costs are forcing tech companies to reconsider their reliance on expensive AI models, potentially shifting workloads to cheaper alternatives.Summary
The AI industry has long operated on the assumption that bigger models are more powerful, but mounting costs are challenging this paradigm. Coinbase co-founder Brian Armstrong predicts that within 12-18 months, 80% of workloads will shift to models that are 99% cheaper, with only 20% requiring the most advanced models. This shift could significantly impact the economics of AI, particularly for major labs like OpenAI and Anthropic, as they face financial pressure ahead of their IPOs. Initial tests, such as those by Harvey, suggest that cheaper models can often match the quality of larger ones when used strategically. The trend highlights a move away from the scaling-first approach, driven by token price increases and reduced investor subsidies, raising questions about the future demand for inference and the justification of training frontier models.
(Source：Tech Crunch)
