On Thursday, Sundar Pichai, the CEO of Alphabet, paid a visit to the White House for a pivotal meeting with high-ranking officials from the Trump administration. The focal point of their discussions was the U.S. government's apprehension regarding the inadequate artificial intelligence computing capacity, a situation that could potentially compromise national defense security reserves. This concern was triggered by the restricted rollout of Anthropic's large-scale model, Claude Mythos, which has caused unease within political circles.

The Trump administration is particularly worried about Anthropic's lack of sufficient computing reserves, fearing that this could hinder its ability to promptly patch software vulnerabilities during times of crisis. To mitigate this risk, the government is keen on harnessing technologies from industry leaders such as Google and OpenAI, with the aim of reducing its dependence on the Claude model.

While Google's in-house TPU chips have proven their worth in many applications, they currently fall short in meeting the stringent requirements of certain secure scenarios. On the other hand, Anthropic's model runs on Amazon Web Services' platform, which has already secured Level 6 security certification. In light of this, the government is expediting the approval process for Google's TPU, striving to ensure it attains equivalent security qualifications in the near future.