Recently, Google has initiated contact with a cluster of small cloud service providers. These providers predominantly source NVIDIA chips for their operations and have been approached by Google to explore the integration of Google's in-house developed chips into their data centers. Insiders with knowledge of the situation reveal that the negotiations have reached a significant milestone: Google has successfully secured an agreement with London-headquartered Fluidstack. Under this agreement, Fluidstack will host Google's Tensor Processing Units (TPUs) in its New York-based data center. Furthermore, Google is currently involved in analogous discussions with other cloud service providers that heavily depend on NVIDIA chips. This strategic move underscores Google's ambition to broaden the adoption of its self-developed chips and potentially intensify direct competition with NVIDIA in the AI chip market.