Andy Konwinski has a blunt warning for American AI companies. The Databricks co-founder and AI investor believes the US is losing its research edge to China, and he's calling it an "existential" threat to democracy.

Speaking at the Cerebral Valley AI Summit this week, Konwinski made a provocative claim. "If you talk to PhD students at Berkeley and Stanford in AI right now, they'll tell you that they've read twice as many interesting AI ideas in the last year that were from Chinese companies than American companies."

That's a striking statement coming from someone deeply embedded in Silicon Valley's AI ecosystem. Konwinski runs Laude, an AI research and venture capital firm he launched last year with NEA veteran Pete Sonsini and Antimatter CEO Andrew Krioukov. He also operates the Laude Institute, an accelerator offering grants to researchers.

The open versus closed debate

Konwinski's argument centres on a fundamental difference in approach. Major American AI labs like OpenAI, Meta, and Anthropic continue innovating significantly. But their innovations remain largely proprietary rather than open source.

Meanwhile, these companies are pulling top academic talent out of universities by offering multimillion dollar salaries that universities simply cannot match. The result? America's academic AI research community is hollowing out.

Konwinski argues this is shortsighted. For ideas to truly flourish, they need to be freely exchanged and discussed within the larger academic community. He pointed to generative AI itself as proof. It emerged directly from the Transformer architecture, a pivotal training technique introduced in a freely available research paper.

"The first nation that makes the next 'Transformer architectural level' breakthrough will have the advantage," Konwinski said.

China's different strategy

In China, according to Konwinski, the government actively supports and encourages AI innovation to be open sourced. Labs like DeepSeek and Alibaba's Qwen release their work publicly, allowing others to build upon it.

This creates a virtuous cycle. More researchers can examine the work, improve upon it, and share their improvements. The result, Konwinski contends, will inevitably lead to more breakthroughs.

He believes this stands in stark contrast to the US, where "the diffusion of scientists talking to scientists that we always have had in the United States, it's dried up."

The business case beyond democracy

Konwinski frames this as both a democratic threat and a business risk to major US AI labs. His metaphor is agricultural. "We're eating our corn seeds; the fountain is drying up. Fast forward five years, the big labs are gonna lose too."

The logic makes sense. If American companies hoard their research while Chinese companies share theirs, the global research community will increasingly build on Chinese foundations rather than American ones. Over time, this compounds. Each generation of research builds on the previous one, and if the best foundational work is coming from China, that's where future breakthroughs will happen.

American companies might maintain short term competitive advantages through proprietary research. But if the broader ecosystem of academic researchers, startups, and international labs is working from Chinese open source foundations, American companies eventually lose their edge.

A self inflicted wound?

What makes Konwinski's argument particularly interesting is that he's not blaming external factors. He's arguing that American AI companies are creating this problem themselves through their closed approach and aggressive talent acquisition from universities.

The irony is thick. American companies are so focused on maintaining competitive moats through proprietary research that they're undermining the open academic ecosystem that produced the breakthroughs they're building on in the first place.

Whether Konwinski is right about the scale of the threat is debatable. American AI labs still produce groundbreaking work, and proprietary research has its place. But his core point is worth considering. If PhD students at Berkeley and Stanford are finding twice as many interesting ideas coming from Chinese companies, something significant is shifting.

His proposed solution is straightforward. "We need to make sure the United States stays number one and open." Whether American AI companies and policymakers will heed that advice remains to be seen.