Sundar Pichai, CEO of Google parent company Alphabet, visited the White House on Thursday and held a series of high-profile key meetings with many senior officials of the Trump administration. According to reports, there is another core issue behind the meeting: the United States is worried about insufficient artificial intelligence computing power production capacity, and the government is unable to reserve sufficient artificial intelligence computing resources to maintain its own national defense and security system.
This concern stems from the Claude Mythos forward-looking large model launched by the artificial intelligence company Anthropic, which is only open to dozens of companies and government agencies. Anthropic said the limited release was based on safety priorities, but the move triggered widespread unease among U.S. political circles and Washington.
Multiple recent tests have shown that this Mythos model can identify and breach security vulnerabilities in critical software infrastructure at a speed that traditional network security teams cannot match.
The Trump administration is worried that Anthropic’s own computing power reserves are insufficient—computing power is an industry term that refers to computing processing power. Even for high-priority institutional users, the company may be forced to limit the scale of Mythos model calls. Some officials worry that in the event of a crisis, the U.S. government may not be able to use the necessary smart tools to patch vulnerabilities in its own software systems.
This also brings opportunities for companies such as Google and OpenAI. Although the government seems to be easing its relationship with Anthropic, it also hopes to reduce its reliance on the Claude model with the help of Google's Gemini large model, OpenAI's GPT series models and other cutting-edge artificial intelligence technologies.
In addition, the U.S. Department of Defense stated last week that it has reached multiple cooperation agreements and will introduce multiple artificial intelligence models in addition to Claude in classified work.
It is understood that Google is currently facing a big problem: its self-developed artificial intelligence processing chip tensor processor (TPU) cannot adapt to some confidential business scenarios. The core reason why Anthropic has become a convenient choice for the government is that most of its models run on the Amazon cloud technology platform; Amazon has invested billions of dollars to obtain the industry's highest level of Level 6 security certification.
The U.S. government is currently seeking ways to accelerate the same security qualification approval process for Google tensor processors.
It is also reported that Anthropic is about to establish a US$1.5 billion joint venture with Wall Street giants such as Goldman Sachs and Blackstone to sell artificial intelligence tool services to private equity holding companies.