Mozilla State of Open Source AI Report (July 14, 2026) — the substantive data anchor

Global

Content

The Mozilla State of Open Source AI Report is Mozilla’s inaugural annual assessment of the open-source AI ecosystem. Published July 14, 2026, based on a global survey of 950+ developers (fielded by SlashData) and new analysis. Provides the substantive quantitative anchors for any talk about open-source AI in agrifood.

Headline findings

• Performance gap with proprietary models: 3%. Open models no longer playing catch-up with ChatGPT, Claude, etc.
• Costs have fallen 50x in three years.
• Open models power ~33% of real-world AI usage but capture only 4% of revenue. Value is real; revenue isn’t flowing back to the open ecosystem.
• China and East Asia lead in open-source AI adoption at 89%, far ahead of the West. Open source is treated as a cornerstone of national strategy.
• 12 new national AI strategies launched last year; 47 countries restrict foreign processing for critical workloads.
• EU, Canada, India back open ecosystems with public investment — treating AI infrastructure as a strategic asset.
• 79% of developers use open models, only 51% have deployed them in production (vs 63% for closed models). The gap is infrastructure, not quality.
• “The real fight has moved beyond the model” — the agentic harness (the software between people and models) matters more than the model itself.
• 93% of users approve AI agent requests by default (“consent fatigue”).

Mozilla’s framing

Raffi Krikorian, Mozilla CTO: “Open source AI has reached a turning point. It’s no longer about expanding access to models; it’s about who has the power to shape, audit, and improve them. Without investment in the infrastructure, tooling, and governance around open models, we risk locking in a system where only restrictive, closed AI can scale – and that doesn’t serve the public interest, or sovereignty over tech policy decisions.”

Mozilla’s “Be Owners, Not Renters” framing: “Open models offer what a subscription never can: owning your infrastructure instead of renting someone else’s. Companies like Microsoft and Uber are already rethinking their reliance on paid, closed AI tools as the bills add up.”

SlashData survey analysis (Álvaro Ruiz Cubero)

“This gap indicates that there is not an issue purely of model quality, but of missing infrastructure. Deployment rates for open models barely increase with company size, highlighting a lack of mature tooling and support. At the same time, buyers are prioritising licensing terms (31%) and ownership (26%), signalling a clear shift toward control and flexibility over raw capability.”

Mozilla’s broader AI work

• Mozilla.ai — open-source libraries for AI agent orchestration, model selection, evaluation, local single-file execution. https://www.mozilla.ai/
• Mozilla Technology Fund — open-source AI for environmental justice (methane emissions tracking, mining operations monitoring, air quality). https://www.mozillafoundation.org/en/blog/open-source-AI-for-environmental-justice/
• AI Manifesto (2025) — open-source AI tools that anyone can review, improve, and customize. Mozilla’s positioning: “rewiring Mozilla to do for AI what we did for the web.”

What this unit is doing in the taxonomy

First claim-type: statistic unit anchored in a major third-party report. Provides the substantive quantitative data anchor for any talk about open-source AI in agrifood. The Mozilla report’s findings are non-vendor; they are independent analysis of the AI ecosystem.

Distinct from:

• Vendor-reported figures (DJI 222M tons water saved, Bayer Climate FieldView 250M acres) — Mozilla is independent.
• Open data ecosystem unit (open-data-ecosystem.md) — about agricultural open data; Mozilla is about AI ecosystem open source.
• Data cooperatives / commons architecture unit (data-commons-architecture.md) — about operational/framework/standards layers of data commons; Mozilla is about open-source AI ecosystem scale and trajectory.

Why it matters for talks

The Mozilla report provides:

• Quantitative anchor — 3% performance gap, 4% revenue capture, 51% deployment rate.
• Substantive framing — “the real fight has moved beyond the model” / “the gap is infrastructure, not quality.”
• Geographic distribution — China/East Asia 89%, West behind.
• Geopolitical context — 47 countries restricting foreign processing; national AI strategies.

These are the load-bearing figures for any talk about open-source AI in agrifood.

Critical context

• The Mozilla report is global AI ecosystem, not agrifood-specific. Application to agrifood requires reading the field guide’s content alongside the Mozilla findings — see open-source-in-agrifood-framework.md for the synthesis.
• The 950+ developer survey is a developer sample, not a farmer sample. The substantive deployment gap finding (51% vs 63%) is about developer deployment, not farm deployment.
• The 4% revenue capture is for open source AI as a category, not specifically open-source AI in agrifood. Worth noting in any talk that cites the figure.
• The China/East Asia 89% adoption figure is for the general AI ecosystem, but it has implications for Chinese agritech AI deployment (DJI, XAG, Alibaba, Pinduoduo) that the China cycle already surfaces.