Nvidia CEO’s Open‑Source AI Crusade: China’s Kimi K3 vs U.S. Policy Tussle

China’s Open‑Source Surge in the U.S. Spotlight

Washington is in a state of panic over open‑source AI systems like Kimi K3 and DeepSeek V4.

Nvidia’s Strategic Defense

In an Axios interview, CEO Jensen Huang stressed that open‑source models such as Kimi K3 are a strong advantage and should not be restricted.

Cost and Competition: Open vs Closed

■Kimi K3: $15 / 1 million tokens

■DeepSeek V4: $0.87 / 1 million tokens

■Anthropic Fable 5: $50 / 1 million tokens

■OpenAI GPT‑5.6 Sol: $50 / 1 million tokens

Open‑source models offer lower usage costs but require powerful hardware and ongoing maintenance. Closed models, while pricier, provide ready‑to‑use convenience.

Policy Reactions and Future Scenarios

■The White House is considering executive powers to impose security conditions and liability acceptance.

■Many U.S. firms are turning to China’s open‑source models for cost savings.

■Huang dismissed the notion of a backdoor, emphasizing that guardrails can be customized.

Markets will react with a mix of optimism for tech growth and caution over geopolitical risks. Open‑source AI could spur Nvidia’s GPU sales and data‑center expansion, but regulatory uncertainties loom. Investors in the Asia‑Pacific should weigh Chinese model pricing and U.S.–China competition in long‑term strategy planning.