Jim Zemlin, CEO, Linux Foundation
Adoption of open source AI now powers every layer of modern computing infrastructure, Linux Foundation Executive Director Jim Zemlin stated during his keynote address at Huawei Connect. Speaking a decade after his first appearance at the event, Zemlin highlighted how shared code has grown from running 80 percent of enterprise systems to underpinning 98 percent of global software applications today.
The Economics of Shared Infrastructure
Zemlin described the current wave of capital expenditure in data centers, energy systems, and specialized hardware as the largest investment in industrial history. This capital wave surpasses earlier historical buildouts in telecommunications, electrical grids, and railroads. Yet, the operational layer running across these facilities relies almost entirely on communal software foundations.
“In a world of competition, open source is a non zero sum game. Nobody’s gain requires anybody’s loss. In fact, a dollar of investment in shared code returns $2,000 in shared value to everybody who builds upon it.”
Jim Zemlin, Executive Director of the Linux Foundation
Citing research from Harvard Business School, Zemlin noted that the collective development model generates trillions of dollars in shared value across the global economy. As infrastructure commoditizes, commercial entities capture real revenue at the application layer through robotics, autonomous mobility, and industrial manufacturing.
How Open Source AI Powers the Agentic Stack
The entire technology pipeline driving agentic workflows depends directly on open projects. Linux and Kubernetes maintain the base infrastructure, while PyTorch serves as the primary engine for training both open and proprietary models. At the inference tier, technologies like vLLM and Ray manage high-throughput serving, while open protocols such as MCP structure autonomous agent communication.
The growth of open source AI at the model layer has reduced per-token processing costs by a factor of ten. Zemlin pointed out that a majority of inference tokens across major routing networks now pass through open weight models. This market shift matches recent public support from 270 technology companies backing open model distribution to prevent monopolistic bottlenecks in artificial intelligence development.
Huawei Contribution and Developer Collaboration
Addressing the host organization, Zemlin commended Huawei for ten years of consistent upstream engineering contributions. He noted that managing shared research and development remains a winning strategy for global enterprises building advanced applications.
Strategic Roadmap for Enterprise Engineering
Organizations looking to lead in next-generation computing must shift from passive consumption to active upstream participation. Zemlin advised engineering teams to commit code directly to foundational projects rather than maintaining isolated internal forks. Contributing directly to shared components allows businesses to redirect their engineering budgets toward proprietary application features where distinct market value is won.
