Amazon CTO says firms are shifting to lower-cost open-source AI

Companies are moving more work to lower-cost open-source AI models as they try to control rising artificial intelligence bills, Amazon chief technology officer Werner Vogels told Fortune. The shift matters for businesses deciding whether they need the most advanced proprietary systems for every task, or whether cheaper tools can deliver enough performance.

Vogels said at the UN’s AI for Good summit that Amazon is seeing customers reassess the balance between less expensive open-source models and larger, costlier systems. He told Fortune that cost has become a key part of how companies design AI systems.

“Do you really need to have the biggest, highest-end model to solve this? The answer is no, you don’t,” Vogels said, according to Fortune.

AI bills are changing buying decisions

Fortune reported that executives have grown more cautious after examples of fast-rising AI costs. Uber said it used its full 2026 AI budget in four months, Fortune reported. Fortune also cited an Axios report that a company spent half a billion dollars in one month after failing to limit employee AI use.

Many frontier AI models from companies including OpenAI, Anthropic and Google DeepMind charge by the token, Fortune reported. A token is the unit of text or data an AI system processes, roughly equal to a word and a half of English text, according to Fortune.

Open-source or open-weight models can often be downloaded without charge, Fortune reported, but companies still must pay for cloud infrastructure to run them. Even with those computing costs, Fortune reported that open-source models can work out cheaper than the most advanced proprietary systems, especially at scale.

Vogels told Fortune that companies are entering a more practical phase of AI adoption after a period of experimentation. Fortune reported that many organizations are now measuring AI projects against return on investment and the cost of running and maintaining them over time.

Transparency is also a factor

Vogels said companies are not looking only at price, according to Fortune. He told the publication that customers also want more clarity about how AI systems are built and trained.

“People want to know what is the data that goes into it,” Vogels said, according to Fortune.

Fortune reported that demand for transparency is especially strong in healthcare, government and humanitarian work, where organizations need to understand how an AI system was trained and how it reaches decisions. Vogels said people serving vulnerable communities need users to trust the systems, or they will not use them, according to Fortune.

Open-source models can let developers inspect and modify code and adapt models to their own data, Fortune reported. Fortune also noted that many open-weight model providers still do not fully disclose all of the data used to train their models.

Amazon adds a research data tool

At the summit, Vogels introduced an Amazon open-source AI tool aimed at helping researchers find scientific datasets more quickly, Fortune reported. The system links the AWS Registry of Open Data, which Fortune said includes more than 1,100 datasets from organizations such as NASA, NOAA and the NIH, to AI assistants.

Fortune reported that the tool lets users search in natural language instead of working through complex data catalogs. It is designed to handle requests such as finding satellite imagery or genomics datasets with specific licensing terms, Fortune reported, with the goal of cutting search times and reducing technical barriers for scientists at under-resourced institutions.

This story draws on original reporting from Fortune.