In a nutshell: Public training data is becoming scarce and expensive, forcing large language model providers to compete for proprietary data and thereby exacerbating market concentration.
The open web as a source for training data is increasingly depleted; high-quality data is becoming scarce, expensive, and increasingly granted exclusively. This does not lead to immediate collapse of AI models, but fundamentally shifts the balance of power in the market.
The era of unlimited availability of publicly accessible training data is coming to an end. Large language models have so far been trained predominantly on freely available text sources from the internet. However, this resource is finite and is being extracted in parallel by established providers, competitors, and newly founded companies.
For Chief Data Officers, this means a strategic reassessment: exclusive or proprietary datasets become a competitive advantage. Organizations must decide whether to use their own data holdings for internal AI training, monetize them, or pursue both paths in parallel. Scarcity is driving prices for high-quality, curated training data upward.
New dependencies are emerging in the market: companies without access to proprietary datasets or without the means for expensive training resources are losing room for maneuver. At the same time, business models around data processing and brokerage are emerging. Power is shifting to those who can control or exclusively license high-quality, relevant training data.
Source: www.golem.de · Published July 2, 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrasing and classification by Lumi News Pipeline v1.7.2.