Google’s Gemini delay exposes a deeper problem: employee morale

Google's Gemini Delay Exposes a Deeper Problem: Employee Morale

Google's most powerful model, Gemini 3.5 Pro, is months late, and a new report reveals the reason extends beyond technical challenges. Low morale, a contentious Pentagon contract, and a talent exodus are hindering DeepMind, despite Google reporting strong earnings and cash flow.

Key Takeaways:

• Morale Issues: Six current and former DeepMind employees share that low morale is impacting model release timelines at the lab.
• Delayed Release: Gemini 3.5 Pro is significantly behind schedule, according to Bloomberg. Google recently released smaller, more affordable models with mixed responses from competitors.
• Talent Exodus: Google has faced significant talent losses in AI, including key figures like Noam Shazeer and John Jumper joining OpenAI and Anthropic, respectively.
• Pentagon Deal Controversy: The April agreement allowing the Pentagon to use Google's technology is a point of disagreement among employees who have resigned over it.
• Google's Defense: Google refutes claims of morale issues and decreased talent retention, citing lower attrition rates in AI staff compared to the previous year and a high acceptance rate for offered AI roles.
• Future Concerns: Despite Google's confidence in its recent model launches and AI safety framework, the company faces potential challenges from rivals like OpenAI and Anthropic.

The Bigger Picture:

While Google can afford to be overtaken on AI leaderboards, it cannot risk losing its top talent or investors' trust.