Famous Developer Reveals Internal Political Struggles in Google's AI Team Led to Abandonment of Gemini CLI Team's Efforts
Famous developer Theo (t3.gg) released a video revealing that internal political struggles within Google's AI team led to the abandonment of the Gemini CLI team's efforts.
After Google hired Windsurf's founder to develop Antigravity, the original team was sidelined. The new product not only frequently consumes tokens with ineffective models but also has a UI that pixel-for-pixel copies Codex, with the demo video even revealing a "Codex" folder.
Theo also harshly criticized Google Cloud for being extremely unreliable, stating that Railway completely crashed due to invisible throttling by Google Cloud, and that the entire UniSuper account was accidentally deleted in 2024, advising developers to migrate out as soon as possible.
Source: Public Information
ABAB AI Insight
Theo is known for his outspoken criticism of major tech companies' engineering cultures. This video continues his previous critiques of Google's product fragmentation and internal incentive mechanisms, having pointed out multiple times that the parallel development of various Gemini versions confuses developers.
Since the launch of the Antigravity project, Google's AI team has shifted core resources from the original CLI and open-source solutions to the new client, motivated by a desire to quickly catch up with Cursor and Codex. However, significant issues of plagiarism and model stability have emerged in execution, while Google Cloud's commercial reliability lags far behind AWS and Azure, with multiple major incidents exposing a lack of operational culture.
Similar to Meta and Microsoft, which have experienced team losses due to prioritizing new projects, and Google's own forced integrations after overlapping product lines, the current state of Google's AI and cloud business is under severe internal political strain and external competitive pressure, increasing the risk of losing talented engineers.
This essentially reflects capital concentration: internal promotions favor "new products," shifting pricing power from a long-term reliable engineering culture to short-term political signals. The mechanism favors projects that can demonstrate quickly rather than maintaining existing assets and infrastructure, leading to the departure of talented individuals and a lack of sincerity in products, ultimately damaging Google's long-term trust and market position in AI agents and cloud services.
ABAB News · Cognitive Law
Internally rewarding only new projects inevitably turns the efforts of old teams into political sacrifices.
Plagiarism can demonstrate quickly, but true sincerity determines whether developers stay.
The less reliable the cloud service, the faster developers leave; reputational collapse occurs faster than technological lag.