Twitter co-founder Jack Dorsey publicly endorsed venture capitalist Chamath Palihapitiya's call to fully embrace open source AI, replying 'yes' to a post that warned US restrictions would leave American firms paying far more for AI than foreign competitors. The exchange comes as Chinese lab Moonshot AI's new model Kimi K3 topped coding benchmarks in July 2026, rattling US chip stocks and highlighting a narrowing capability gap between Chinese and American AI systems.
Why the cost gap matters
Palihapitiya laid out stark numbers: US restrictions on open source AI would force American companies to pay $26 to $56 per million tokens for AI, while foreign rivals can buy the same for $0.50 to $1. He called the gap unsustainable. He extended the argument to national defense, saying that paying high costs to defend US systems while adversaries attack cheaply creates a similar imbalance. The post drew hundreds of thousands of views within hours, indicating the debate resonates beyond Silicon Valley.
Chinese AI advances and US policy
Moonshot AI's Kimi K3 model topping coding benchmarks in July 2026 sent ripples through US chip stocks. David Sacks echoed Palihapitiya's view, agreeing with researcher Sebastian Mallaby that dangerous AI capability will soon spread freely regardless of policy. Sacks cited concerns around Anthropic's Claude Mythos model. He predicted Chinese models would reach advanced cyber capability within months and noted that Washington's staggered release of GPT-5.6 over security worries failed to slow foreign progress.
Dorsey's endorsement and the open source push
Dorsey's endorsement carries weight given Block's own open source AI agent Goose, which he has championed publicly for years. His simple 'yes' reply to Palihapitiya's post signals alignment with a growing faction that sees open source as the only way to keep US AI competitive. The post's rapid spread suggests the issue is gaining traction beyond tech insiders.
The policy debate
US policymakers are debating how tightly to control advanced AI models. Proposals include vetting models before release to manage security risk without ceding ground to China. But Palihapitiya and Sacks argue that restrictions only hurt American firms while foreign rivals — especially in China — continue to advance. The question now is whether Washington will tighten controls or shift toward the open source approach Dorsey and Palihapitiya advocate.
