Box Inc. CEO Aaron Levie warned that recent developments in AI governance could accelerate a global shift toward open-weight artificial intelligence models as countries seek greater control over their own digital infrastructure. Levie argued that uncertainty around AI access is pushing nations to reconsider reliance on U.S.-controlled systems, introducing strategic risk for governments and businesses dependent on foreign-built AI.

Open-Weight AI Models And Sovereign AI Shift

In a post on X on Sunday, Levie stated that the "big winner" in the current environment would be open-weight models. He noted that a new precedent has emerged regarding the possibility that "a model could be pulled back," creating a dynamic that introduces strategic risk for entities relying on foreign technology. He added that if access can be restricted at any time, "this poses very real risk on relying on technology from a particular country."

Levie suggested that regulatory focus on the "model layer" rather than the "applied layer" could have unintended geopolitical consequences. He wrote that the game theory the U.S. should consider is that other countries now have increased incentive to develop sovereign AI. Consequently, the most likely response from other nations would be increased adoption of open-weight models, which allow independent deployment and customization. Levie noted these systems are "generally not coming from the US" at present.

AI Scrutiny And Industry Developments

The comments arrive amid broader regulatory scrutiny in the AI sector. OpenAI was previously investigated by multiple state attorneys general over its data practices, safety measures, and consumer impacts ahead of a potential IPO. The company stated it would cooperate with regulators.

Separately, NVIDIA Corp. announced a $26 billion plan to develop open-weight AI models, a move supported by Block Inc. CEO Jack Dorsey. This expansion signals a shift beyond chips into model development, driven by global competition and concerns about technological fragmentation. Meanwhile, Anthropic CEO Dario Amodei has warned that advanced AI systems require stricter oversight, arguing that governments should be able to block or reverse unsafe deployments.