American AI companies can’t get enough chips

In 2026, artificial intelligence (AI) chip production has become a binding constraint on the pace of the AI compute buildout. Demand for computing power to train and deploy advanced AI models continues to grow exponentially, outpacing many chip manufacturers’ forecasts.

Supply chains for AI chips and key inputs cannot scale rapidly enough to meet demand, as it takes years to build additional manufacturing capacity. Given these constraints on AI chip supply, the United States has both greater leverage and greater reason to ensure every chip is put to its highest-value use, giving rise to five key policy implications:

• As rising chip costs risk pricing out researchers and scientists, Congress should significantly increase funding for the National AI Research Resource to ensure compute access for public innovation keeps pace.
• Every chip exported to competing countries such as China is one fewer available to U.S. companies and democratic allies. Exports to China expand the compute accessible to the Chinese Communist Party (CCP) that can be used against American interests.
• Aggressively exporting AI chips to allies and democratic partners remains critical to enable U.S. AI companies to procure sufficient data center capacity and to ensure long-term American AI leadership.
• Chip shortages add urgency to countering chip smuggling. Location verification and stronger controls on high-bandwidth memory can help ensure scarce supplies are not diverted to unauthorized end users.
• New initiatives like Pax Silica have a vital role to play in coordinating efforts between allies, helping build out resilient semiconductor supply chains.

Read more at Center for a New American Security