A leading brokerage recently set a target price of $390 for Google, and the underlying logic is worth analyzing.

First, let's look at two key actions: one is securing NATO's AI cloud project - this is not an ordinary client; it means it has passed military-level scrutiny in terms of data security and cross-border compliance. The system design is quite clever: sensitive information is physically isolated within the country, but AI analysis capabilities can still be accessed, satisfying sovereignty requirements while also unleashing computational power value.

Another is the Meta order that is currently under discussion, which could reach tens of billions of dollars. If Meta really hands over its AI training needs to Google's TPU chips, it essentially acknowledges a fact: in the high-performance computing race, Google's hardware ecosystem is already able to compete with Nvidia.

The brokerage raised the valuation from 360 to 390, with the core judgment being that the market has underestimated several hidden lines: the enterprise-level penetration rate of Cloud business, the bargaining power of TPU, the data barriers of Waymo's autonomous driving, and the AI monetization capabilities hidden within SaaS.

Looking at it from another angle, Google's current state is somewhat similar to Amazon's situation ten years ago when it had just developed AWS—on the surface, it appears to be a tech company, but in reality, it is transforming into an infrastructure provider. The difference is that this time, the infrastructure is not cloud storage, but AI computing power scheduling and model services.

The essence of this round of valuation reassessment is actually pricing the "Year of AI Commercialization."