Major AI Firms Pay for Premium Wikipedia Data Access Amid Rising Infrastructure Costs

The Wikimedia Foundation announces new API access deals with major AI firms, including Microsoft, Meta, Amazon, Perplexity, and Mistral AI, to provide high-volume, high-speed access to Wikipedia's content. These companies, which use the data to train AI models like Microsoft Copilot and ChatGPT, now join the foundation’s Wikimedia Enterprise program, a commercial subsidiary that offers faster and more extensive data access than the free public APIs.

Expanding the Wikimedia Enterprise Program

The new agreements mean that most leading AI developers are now part of the Wikimedia Enterprise program, which sells high-speed API access to Wikipedia’s 65 million articles. While Wikipedia’s content remains freely available under a Creative Commons license, the Enterprise program charges for the premium access. The financial terms of the deals remain undisclosed.

Google, which signed a deal with Wikimedia Enterprise in 2022, is already a partner, along with smaller companies like Ecosia, Nomic, Pleias, ProRata, and Reef Media. The revenue from these partnerships helps offset the infrastructure costs for the nonprofit, which otherwise relies on small public donations.

Addressing Infrastructure Costs and Scraping Issues

The push for paid API access comes after years of rising infrastructure costs due to industrial-scale scraping by AI companies. In April 2025, the foundation reports a 50 percent increase in bandwidth used for downloading multimedia content since January 2024, with bots accounting for 65 percent of the most expensive requests to core infrastructure, despite making up just 35 percent of total pageviews.

By October, the Wikimedia Foundation discloses that human traffic to Wikipedia has fallen approximately 8 percent year over year. This decline threatens the feedback loop that has sustained Wikipedia for a quarter century: readers visit, some become editors or donors, and the content improves. However, many AI chatbots and search engine summaries now answer questions using Wikipedia content without sending users to the site itself.

Resistance and Support for AI Integration

The foundation’s own experiments with generative AI have met resistance from volunteer editors. In June, Wikipedia pauses a pilot program for AI-generated article summaries after editors call it a “ghastly idea” and warn it could undermine trust in the platform. Wikipedia founder Jimmy Wales supports AI models training on Wikipedia data, stating, “I’m very happy personally that AI models are training on Wikipedia data because it’s human curated.” However, he emphasizes the need for fair compensation: “You should probably chip in and pay for your fair share of the cost that you’re putting on us.”

“Wikipedia is a critical component of these tech companies’ work that they need to figure out how to support financially,” says Lane Becker, president of Wikimedia Enterprise. “It took us a little while to understand the right set of features and functionality to offer if we’re going to move these companies from our free platform to a commercial platform… but all our Big Tech partners really see the need for them to commit to sustaining Wikipedia’s work.”

References

• Wikipedia signs major AI firms to new priority data access deals - Ars Technica