US lawmakers have expanded a joint congressional investigation into American technology companies and startups evaluating Chinese artificial intelligence software.
The House Select Committee on China and the House Committee on Homeland Security sent a formal inquiry to food delivery giant DoorDash regarding its integration of open-weight models developed by Beijing-based startup Moonshot AI.
According to the House Select Committee official statement, the probe targets growing national security, cybersecurity, and economic implications as US businesses integrate foreign open-source architecture into core operations. DoorDash is the third major US firm brought under congressional inquiry following earlier letters sent to travel platform Airbnb and AI coding startup Anysphere.
Why Startups Are Turning to Chinese Open Weight Models
The biggest reason behind this shift is simple. Many closed-source AI providers still charge high API prices, making it expensive for startups and even large consumer platforms to build and scale AI products. As AI usage grows, these computing costs increase quickly.
Open-weight models from Chinese AI labs, including Moonshot AI’s Kimi series, offer comparable performance benchmarks at a fraction of the operating cost. Because open-weight models publish dictating neural weights publicly, engineering teams can download, self-host, and fine-tune software locally without sending user data back to external servers.
Nvidia CEO Jensen Huang has also defended the use of high-quality open-weight AI models regardless of where they are developed.
Speaking about Chinese AI models, Huang said, “These Chinese models are excellent. Open-source models that are excellent should be used.” He has argued that restricting access to competitive AI technology could ultimately slow innovation in the United States.
As reported by the South China Morning Post, congressional leaders acknowledged in detail that startups are evaluating these open-weight alternatives because they provide competitive reasoning capabilities, greater customisation, and relief from reliance on a small oligopoly of domestic model suppliers.
National Security Scrutiny and the Threat of Compliance Friction
Despite commercial benefits, Washington officials warn that deploying models subject to foreign jurisdiction presents structural risks. Lawmakers also argue that open-weight architectures lack standardised guardrails, creating vulnerabilities to code tampering or hidden software backdoors.
As noted in a legal analysis by VitalLaw, committee chairmen have requested detailed infrastructure records, security evaluations, and executive briefings from corporate leadership. This escalating oversight signals a sharp increase in regulatory scrutiny for startups.
Founders who adopt foreign open-weight models purely to reduce burn rate may inadvertently inherit geopolitical friction, potential supply chain restrictions, and heightened due diligence requirements from institutional investors and corporate clients.
Building a Resilient Multi Model Strategy for Founders
Startups can also take an important lesson from this situation. Depending on just one AI model because it is cheaper can become a problem if regulations change or the provider faces restrictions. It is safer to build products that can easily switch between different AI models without major development work.
If your startup uses open-weight models, don’t ignore security. Regular security testing, code reviews, and strong data protection should be part of your development process from day one. Enterprise customers want affordable AI, but they also expect clear proof that their data and systems are secure.
