White House Approves Secret $9 billion AI Chip Push for US Spy Agencies
The White House has reportedly approved a classified $9 billion request to acquire advanced computer chips for US intelligence agencies. With this, the aim is to help them fully utilize next-generation artificial intelligence (AI) systems amid growing concerns over a shortage of computing power.
Citing current and former US officials, The New York Times reported that the funding push reflects mounting concerns inside the administration and Congress that agencies such as the Central Intelligence Agency (CIA) and National Security Agency (NSA) risk falling behind in testing and deploying cutting-edge AI tools for classified intelligence work.
The proposed funding, which still requires congressional approval, is designed in part to support infrastructure capable of running Nvidia’s Grace Blackwell superchips, processors considered essential for advanced AI models but requiring specialized, power-intensive data centers with liquid cooling systems.
Meanwhile, the administration is reportedly redirecting $800 million to accelerate access to computing capacity.
AI is Increasingly Central to National Security
AI has become increasingly embedded in military and intelligence operations, helping agencies process massive volumes of classified data and identify overlooked intelligence signals, including communications intercepts.
“Our intelligence community needs the frontier, the best AI chips, models, systems, talent, on a timeline that matches the threat,” Vinh Nguyen, former NSA chief data scientist and current senior fellow on AI at the Council on Foreign Relations, told The New York Times.
While the chip shortage has reportedly forced workarounds, White House Chief of Staff Susie Wiles reportedly authorized the NSA to continue using an advanced model from Anthropic, despite the Pentagon reportedly labeling the AI company a supply-chain concern.
US officials told the newspaper that Anthropic and the government are finalizing a classified agreement to maintain NSA access to the company’s technology.
Anthropic’s Mythos Model Under Scrutiny
At the center of the discussions is Anthropic’s Mythos AI model, released in April and reportedly capable of identifying and exploiting cybersecurity vulnerabilities at an advanced level.
Due to security concerns, the model has only been shared with a limited group of government agencies, banks and firms in the US and Britain.
“Sensitive national security deliberations are conducted with the seriousness they demand, not leaked to reporters,” White House spokesperson Steven Cheung told The New York Times, dismissing what he described as “unverified claims”.
AI for National Security
The report comes as the Trump administration intensifies its focus on AI, including efforts to create frameworks for sharing frontier AI systems with national security agencies before public release.
However, as reported earlier by IBTimes Singapore, Trump has postponed the signing of an executive order on AI, saying he had concerns that parts of the proposal could undermine the US’ advantage in the global AI race, particularly against China.
The executive order was expected to be signed at a White House ceremony, attended by top executives from leading AI companies. However, Trump told reporters in the Oval Office that the event was delayed after he raised objections to certain provisions in the draft.
Additionally, it was also reported that the US War Department announced the signing of new agreements with eight leading tech companies to deploy advanced next-gen capabilities across its classified networks.
The said agreements involve major technology players, including SpaceX, OpenAI, Google, NVIDIA, Reflection, Microsoft, Amazon Web Services (AWS) and Oracle.
