Andrew Feldman, CEO of AI chip company Cerebras, warns that the United States is falling behind China in the artificial intelligence race, attributing the gap to the US’s decentralized government and its patchwork of regulations.
Feldman argues that China’s central decision-making allows it to rapidly develop the power infrastructure and data centers essential for large-scale AI deployment. China has built extensive power infrastructure—including massive dams, coal plants, and solar investments—benefiting from a national-level plan.
In contrast, the US system, with its layers of local, state, and federal rules, creates significant obstacles. “Our decentralized form of government has left us with sort of a patchwork of power infrastructures,” Feldman told Business Insider. He noted that local city and county ordinances can interfere with and set back multi-billion-dollar infrastructure projects, even when the federal government supports them.
Fragmentation as a ‘Tax on Innovation’
Feldman criticized the political fragmentation, stating, “There is no reason why we should have local ordinances interfering with the development of power projects… and the development of data centers.” He also called the current legislative environment a “tax on innovation,” pointing out that a patchwork of state-level AI regulations forces companies like Cerebras and OpenAI to navigate different rules in every state.
While acknowledging that the US shouldn’t simply copy China, Feldman stressed the need for better ways to support domestic AI investment through the existing system.
Proposal for a State-Level Regulatory Freeze
To help the industry, Feldman suggested a moratorium on state-level AI laws. He proposed an initial five-year freeze, with an option to renew, arguing that this would keep pace with the market’s “unbelievably quickly” movement and benefit both large and small AI companies. This idea echoes an unsuccessful attempt by Senator Ted Cruz to block state and local AI laws for a decade.
As a player in the AI race, Cerebras is developing AI training chips. The company recently withdrew its formal filing to go public but stated its intention to update the filing and pursue an IPO in the future.
