虎嗅

Four top Google executives have left to start their own businesses, and their former employer is providing both funding and computational resources: How can companies truly retain their best talent?

原文:谷歌四位大神离职创业,老东家出钱又出算力:怎样才算真正留住人才?

Summary of Key Points

Google has recently made a significant move: there have been personnel adjustments in its AI department and several top talents have left to start their own businesses, yet they haven’t completely severed ties with the company. Demis Hassabis has shifted from managing DeepMind’s day-to-day operations to focusing on long-term research for AGI (Artificial General Intelligence), while David Corray is in charge of implementing the company’s products. Four other prominent individuals, including Jeff Dean, have left to establish new companies, but Google has become their investor, providing them with computing power and cloud services. This isn’t just a simple case of “talent loss”; rather, it represents Google moving some of its cutting-edge explorations outside the company’s boundaries. By investing in and collaborating with these individuals, Google is able to maintain focus on its main project, Gemini, while also allowing for independent exploration.

Detailed Analysis

1. It’s Not About “Talent Leaving,” but About Moving Exploration Outside the Company

When people like Jeff Dean leave to start businesses, Google doesn’t react negatively; instead, it provides them with funding, computing resources, and becomes a strategic partner. It’s like when your child wants to open their own shop—you don’t stop them; you offer support and resources so they can succeed. By doing this, Google allows explorations that might have been restricted within the company (such as Dean’s idea of automating AI experiments) to proceed freely, while still benefiting from the results.

2. Google Has Changed: From “Holding Employees” to “Connecting with an Ecosystem”

Previously, Google aimed to keep all top talents as employees. However, it realized that forcing them to stay (only to potentially join competitors) wasn’t effective. Now, by investing in and collaborating with these individuals, it keeps them within its “circle of influence.” For example, Dean’s new company uses Google’s cloud services, allowing Google to gain valuable feedback on their technology. This approach expands the company’s reach—people don’t necessarily have to be employees; as long as they are part of this ecosystem, they become part of Google’s community.

3. Top Talents Care More About Control Than Money

The reason these individuals leave is not about the amount of money offered; rather, it’s about having the power to make decisions. For instance, Hassabis wanted to develop high-frequency trading algorithms but was rejected by Google, and DeepMind’s attempt to establish an independent board of directors was unsuccessful. What top talents want is the freedom to pursue their ideas with sufficient funding, computing resources, and access to the right people without going through lengthy approval processes. By letting them start their own businesses, Google gives them that control while maintaining a connection through investment.

4. Focusing on the Main Project While Allowing for Independent Explorations

By merging DeepMind and Google Brain, Google accelerated its progress on Gemini (to compete with ChatGPT). However, this also meant that different ideas had nowhere to go. The new approach allows for internal focus on Gemini (with Corray in charge) while external partners like Dean can explore uncharted territories (such as AI experiments). This ensures both the main project moves forward quickly and opens up possibilities for future innovations.

5. Lessons for Other Companies

Google’s strategy can be useful for ordinary businesses:

  • Separate Task Types: Manage routine tasks (like product iterations) centrally, while giving small teams with limited budgets time to test unproven ideas (new business directions) for three months without linking them to performance evaluations.
  • Reduce Redundant Approval Processes: Assess how long it takes from an idea to an experiment and minimize unnecessary approvals. In the AI era, slow approval processes waste both resources and opportunities.
  • Leaving Employees Isn’t a Problem: If someone wants to start a business, don’t call them a “traitor”; instead, consider collaborating or investing to maintain a potential connection in the future.
  • Freedom with Limits: Give freedom to explore, but set clear boundaries regarding budget, time, and termination criteria—don’t indulge without any limits.

Final Thoughts

Whether Google will fail is still uncertain. However, this approach of combining internal focus with an external ecosystem could be a new way for large companies to retain talent and maintain innovation. After all, top talents want the freedom to do what they want while having access to the necessary resources. Other companies should consider this: don’t confine their talents within rigid organizational structures; let them thrive within your ecosystem, as it’s often more effective than trying to keep them tightly controlled.