databricks co-founder argues us must go open source to beat china in ai

The U.S. Losing Ground in AI Research?

Andy Konwinski has voiced concerns regarding a potential decline in the United States’ leading position in AI research, attributing this shift to advancements originating from China. He characterizes this development as a significant, even “existential,” challenge to democratic principles.

Konwinski, a co-founder of both Databricks and the AI research and venture capital firm Laude, shared his observations. He noted a changing perception among doctoral students at prominent institutions like Berkeley and Stanford.

Shifting Perspectives of AI Researchers

According to Konwinski, current PhD students in AI report encountering a greater volume of compelling AI concepts in the past year stemming from Chinese enterprises than from their American counterparts.

Beyond his involvement with Laude, where he invests alongside Pete Sonsini and Andrew Krioukov, Konwinski also directs the Laude Institute. This institute functions as an accelerator, providing financial grants to researchers in the field.

The Impact of Proprietary Research

While major AI laboratories, including OpenAI, Meta, and Anthropic, continue to drive innovation, a substantial portion of their work remains within a proprietary framework, rather than being openly accessible.

These companies are also attracting highly skilled academic professionals by offering exceptionally high salaries, often exceeding the earning potential available within university settings.

The Importance of Open Exchange

Konwinski emphasizes that the free exchange and discussion of ideas within the broader academic community are crucial for genuine progress. He highlights the emergence of generative AI as a direct consequence of the Transformer architecture.

This pivotal training technique was initially introduced through a publicly available research paper, demonstrating the power of open collaboration.

The Race for the Next Breakthrough

“The first nation to achieve the next breakthrough at the level of the ‘Transformer architecture’ will gain a considerable advantage,” Konwinski stated. This underscores the importance of continued innovation and accessibility.

China's Approach to AI Development

Konwinski contends that the Chinese government actively supports and promotes the open-sourcing of AI innovations. This includes work from labs such as DeepSeek and Alibaba’s Qwen, fostering a collaborative environment.

He believes this approach will inevitably accelerate further breakthroughs as others build upon existing foundations.

A Decline in Scientific Collaboration in the U.S.

In contrast, Konwinski observes a diminishing level of interaction and knowledge sharing among scientists within the United States. He suggests that “the diffusion of scientists talking to scientists…it’s dried up.”

Long-Term Consequences

Konwinski warns that this trend presents not only a risk to democratic ideals but also a potential business threat to leading U.S. AI labs. He expressed concern that current practices are unsustainable.

“We’re eating our corn seeds; the fountain is drying up,” he cautioned. “Fast-forward five years, the big labs are gonna lose too.” He advocates for maintaining U.S. leadership and fostering an open AI ecosystem.