Self-developed chips: It's not just about "substitution," but a "profit defense war"
• Directly targeting Nvidia's GPUs (such as H200/B200)
• Only when paired with Google's self-developed Jupiter high-speed network and liquid-cooled data centers can the cluster's power be fully unleashed
The key to its output (ROI) isn't "how much money is made from selling chips," but rather "how much is saved on buying graphics cards"
• Currently, Google has 950 million monthly active users for Gemini. If all used Nvidia chips, profits would be eaten up by hardware costs. Self-developed TPUs lower computing power costs, leading to direct gross margin improvement
There are also key changes in TPU's revenue model
• The CFO explicitly mentioned that the surge in backlog orders was partly due to the inclusion of "TPU hardware sales"
Here it comes, the first day of the earnings explosion
The financial calendar analysis agent mentioned last time that it could provide post-earnings interpretation, so I quickly asked it about the latest earnings report for $Alphabet - C(GOOG.US)
Three key highlights:
1. Google Cloud accelerating its boom (YoY +82%) 👉 Strong demand for enterprise AI infrastructure
2. AI tools landing — Antigravity users surging vs. Gemini 3.5 Pro delayed [Risks cannot be ignored]
3. Capital expenditures hit a new high again (full-year guidance $195-205 billion) 👉 Backing the upstream supply chain
The copyright of this article belongs to the original author/organization.
The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.
