Kei Koizumi, a former senior White House Office of Science and Technology Policy official and a key architect of the 2022 Chips and Science Act, says the law’s manufacturing goals were implemented quickly under the Biden administration but have since been weakened. He points to canceled or modified programs, reduced commitment to flexible implementation, and proposed cuts to research funding as signs that the current administration has undercut the law’s broader promise. He argues that semiconductor strategy requires sustained government attention beyond passing legislation, especially on workforce development, research capacity, and the long-term execution needed to support projects such as domestic fabrication plants.
Koizumi says the United States has struggled to maintain a consistent industrial policy because political cycles make long-term planning difficult for companies. He describes the CHIPS effort as a major experiment in civilian industrial policy after decades in which the U.S. mainly sustained long-term commitments in military technologies. He is sharply critical of newer forms of intervention, particularly the government taking an equity stake in Intel, saying such moves resemble state ownership that has historically not been the U.S. model. In his view, China’s advantage lies in its unified government approach and its ability to align state, military, and commercial priorities, even if that model also carries risks.
On semiconductor exports, Koizumi criticizes the Trump administration’s decision to allow Nvidia to export H200 Artificial Intelligence chips to China, arguing that it bypassed the case-by-case national security process built into export controls. He says the administration may ultimately prove correct in trying to keep China dependent on the U.S. tech stack, but contends that it has not adequately balanced economic aims with military and security risks. His warning is blunt: access to top-tier chips could help China strengthen military Artificial Intelligence capabilities and improve generative Artificial Intelligence systems that shape information and behavior. He also says Nvidia’s influence in Washington now appears unusually direct, describing the current environment as a troubling form of crony capitalism.
Koizumi sees enduring U.S. strengths in Silicon Valley, venture capital, and decades of federal investment in fundamental research. But he says those advantages are now at risk as China expands its own support for Artificial Intelligence research and as Washington considers deep cuts to scientific funding. He says the United States does not need to lead every field, but it must remain among the leaders across key technologies or risk losing the research base that supports venture capital and industrial leadership. He also defends continued scientific collaboration with China under the renewed Science and Technology Agreement, while arguing that some fields, especially Artificial Intelligence, require a high burden of proof and proper security safeguards. His top policy message is that continued research investment and guardrails for advanced technologies remain essential to protecting U.S. interests.
