Revenue Threshold Creates New Licensing Precedent
Chinese AI startup Z.ai has made the weights for its latest open model, GLM-5.3, publicly available under a new licensing agreement that imposes restrictions on large corporations. The release, announced on August 28, 2026, applies specifically to companies generating more than $10 billion in annual revenue over a 12-month period.
Breaking news
Eufy Unveils Local AI Home Security Ecosystem at IFA
The Rapid Evolution of Data Center Security in the AI Era
The High-Voltage Risks Facing Modern AI Data Centers
Apple’s New CEO Renames Lake Ontario To Lake America In Maps AppThe licensing framework represents a departure from traditional open-source models, introducing a tiered access system based on commercial scale. Under the terms, qualifying high-revenue companies must meet additional requirements before deploying the model at scale. Z.ai did not specify the exact nature of these obligations, but the move signals growing tension between open innovation and commercial dominance in the AI sector.
The $10 billion revenue cutoff targets a narrow group of global tech giants, including firms like Amazon, Google, Microsoft, and others operating across multiple sectors. This threshold effectively excludes most startups and mid-sized enterprises from the restricted category, allowing them broader access to the model weights. Industry analysts suggest this approach could influence future open-model releases, encouraging other developers to adopt similar revenue-based licensing structures.
How Will This Affect Enterprise Adoption?
Z.ai’s decision aligns with broader trends in AI governance, where creators seek to balance openness with control over how their innovations are used. By limiting unrestricted access among the largest players, the company may be attempting to prevent its technology from being absorbed into proprietary ecosystems without reciprocal contribution.
The licensing change introduces uncertainty for major cloud providers and enterprise software vendors that rely heavily on open models for product development. Companies falling above the revenue threshold will need to negotiate separate agreements or explore alternative models, potentially slowing deployment timelines. Meanwhile, smaller organizations and academic institutions retain relatively unfettered access, which could shift innovation momentum away from dominant platforms.
Legal experts note that such revenue-linked licenses walk a fine line between enforceability and practicality. While not unprecedented, they remain untested in court and may face challenges regarding jurisdiction and compliance across international borders.
Looking ahead, Z.ai’s licensing model may prompt renewed debate about what constitutes true „openness” in AI. If widely adopted, it could fragment the open-model ecosystem and reshape how foundational technologies are shared and monetized globally.
Frequently Asked Questions
What defines a qualifying company under the new license? Companies with over $10 billion in revenue within any 12-month period are subject to additional terms when using GLM-5.3 weights.
Can smaller firms use the model freely? Yes, organizations below the revenue threshold can access and deploy the model without extra restrictions.
Is this type of licensing legally binding? Revenue-based licenses are novel and lack extensive legal precedent, raising questions about enforceability across jurisdictions.
