US spy agencies race for Nvidia AI chips under new $9 billion plan

The White House has officially approved a $9 billion funding initiative aimed at helping the US intelligence agencies to rapidly adopt and integrate advanced artificial intelligence capabilities. The request was made to acquire cutting-edge computer chips that American spy agencies needed to boost model capabilities. The $9 billion is in part intended to use and increase Nvidia’s Grace Blackwell superchip, which requires enormous amounts of electrical energy and specialised liquid cooling systems. It is now subject to the approval of Congress.

The US intelligence community is facing a serious shortage of advanced AI chips needed for next-generation surveillance, cyberwarfare, and intelligence analysis. Not just the intelligence community, the entire AI stack is facing an extended demand for compute power and semiconductor manufacturing capacity. US intelligence agencies like the National Security Agency and the Department of Defence fear they are falling behind private AI labs and geopolitical rivals. The Trump administration is considering allocating $800 million for a more rapid acquisition of computing capacity

“Our intelligence community needs the frontier - the best AI chips, models, systems, talent - on a timeline that matches the threat,” said Vinh Nguyen, the former chief data scientist at the NSA and a senior fellow on AI at the Council on Foreign Relations.

Also Read: Trump says he will speak to Taiwan President amid arms sale tensions with China

US is treating AI like a national security asset

The US government has started to consider AI as infrastructure as a national security asset comparable to nuclear technology or satellite systems. The funding aligns with several other recent national security steps taken by the administration. The US government agencies run their AI models on Amazon Web Services, which has recently announced $50 billion for upgrading its government cloud computing services.

Top tech firms OpenAI, Anthropic, and xAI have agreed to allow pre-release access to advanced AI models for government vetting. While Anthropic appeared more cautious and tried to maintain restrictions on military and surveillance usage, which reportedly led to tensions with the Pentagon earlier in 2026. The two sides had a clash when Anthropic resisted the Defence Department's demand for the authority to employ Anthropic’s technology for “any lawful use.” To work around the chip shortage, Susie Wiles, the White House chief of staff, has authorised the National Security Agency to continue to use an advanced model made by Anthropic despite the Pentagon designating the company a supply chain threat. The NSA is now finalising a contract to keep using the Mythos model of Anthropic.

Trump was expected to sign a new AI executive order in May 2026, but abruptly cancelled the signing just hours before the ceremony. The draft was a way to regulate the AI models, a precursor to licensing. But it was cancelled after pressure from Elon Musk, Mark Zuckerberg and former AI adviser David Sacks. Trump argued he didn't like “aspects of it”. Trump himself has said before, “We’re leading China … I don’t want to do anything that’s going to get in the way of that lead.” Donald Trump previously revoked a major AI safety executive order issued by Joe Biden shortly after returning to office in January 2025.