Key Takeaways

  • China is reportedly considering restricting overseas access to its top AI models, viewing research leaks as a national security issue, echoing a tactic of going open-source to catch up, then closing models to capture value.
  • This mirrors David Sacks' observation of OpenAI's strategy, suggesting that early openness in AI often gives way to proprietary closure once market leadership is established.
  • Brad Gerstner believes China's move won't harm the US, citing Washington's unified bipartisan goal to maintain AI dominance and the capacity for the US to develop its own open-source alternatives.
  • Every nation, per Chamath Palihapitiya, is now actively pursuing a "sovereign AI" strategy, aiming for independent AI stacks rather than relying on foreign, closed-source models.
  • David Sacks predicts the AI market could consolidate into a duopoly of top labs, leaving everyone else behind, a trend founders must consider for long-term viability.

China's AI Playbook: Open Source to Lock-In

Last week, the chatter on the All-In Podcast turned sharp when Jason Calacanis cited a Reuters report: “CCP officials, Chinese Communist Party, are reportedly considering restricting overseas access to China's top models.” This isn't just about data security; it's a strategic move in the global AI race, one with a familiar rhythm. David Sacks quickly framed it as a classic "open then close" maneuver. “I think the tactic is you stay open until you catch the frontier or you get close to it and then there's a really compelling incentive to go close because you want to capture all the value for yourself,” Sacks explained. He pointed out that this playbook isn't unique to nation-states; it's the exact path OpenAI has walked, moving from an open-source genesis to a highly proprietary, closed-model future once they had a lead.

For founders, this shift isn't a theoretical exercise. It’s a warning. The current era of relatively accessible, powerful AI models—often with generous free tiers or low-cost APIs—might be a fleeting window. If leading models, whether from China or the US, increasingly close their gates or exert more control, startups building on these platforms face mounting dependency risk. Your entire product could rely on a black box that changes its pricing, access rules, or even availability with little notice. This isn't just a concern for foreign policy wonks; it's an immediate, practical threat to your product roadmap and business model.

The Global Scramble for Sovereign AI

While China weighs its restrictions, the world isn't sitting still. Chamath Palihapitiya underscored a broader truth: “There is not a single country in the world that is not trying to figure out its own sovereign AI strategy. And I don't think they believe using a closed source American model is the answer.” This means that beyond the US-China dynamic, every nation from Germany to India is eyeing its own independent AI stack—local data, local compute, locally controlled models. They want to avoid a future where their national security, economic competitiveness, or even cultural identity is dependent on foreign tech giants.

Brad Gerstner, however, expressed confidence in the US position. He argued that even if China restricts access to its models, it wouldn't harm the US, which can develop its own open-source alternatives. Gerstner also noted the unusual unity in Washington: “The one thing there's absolute agreement on is doing everything to stay ahead of China... It is a unifying force in Washington.” This bipartisan focus means significant resources will continue to pour into US AI research, both open and closed source. But this isn't a global win-win; it’s a fragmentation. The era of universal AI tools is ending, replaced by national digital borders and competing technological ecosystems. Your global expansion strategy just got a lot more complicated.

AI's Inevitable Duopoly?

Further complicating this fragmented future is David Sacks' stark prediction about market consolidation. “A year ago it seemed like we had five major labs you know now it seems like there's a top two and then everybody else. So I mean look I could see AI easily becoming another tech market that becomes a duopoly,” Sacks observed. If the frontier AI space does narrow to just two dominant players, the implications are massive. For nations pursuing sovereign AI, building an independent stack becomes even harder if only two entities hold the keys to cutting-edge models.

For founders, this means the competitive landscape will condense dramatically. You'll likely be building on, or competing with, the output of these top two labs. The "open then close" strategy combined with a duopolistic market means the window for truly disruptive, independent foundational AI research outside these top players (or national efforts) may be closing fast. This is not a market for the faint of heart; it's a battleground where only the most strategically astute, or politically aligned, will thrive.

What to Do With This

First, audit your AI model dependencies immediately. If your product relies on a single closed-source model, begin prototyping with open-source alternatives or considering multi-model strategies. Second, rethink your international expansion: if every nation wants its own "sovereign AI" stack, generic global products will struggle. Instead, explore opportunities in localization, building specific applications for national AI initiatives, or addressing unique regional data needs. Finally, if you're building a foundational AI model, understand that the long-term game is likely about going open to gather adoption, then closing to capture value—but be prepared to navigate a world increasingly fragmented by national interests and a market dominated by just a few giants.