AI Geopolitics: Google Former CEO on China Strategic Open-Source Play with DeepSeek

Updated: May 06 2025 17:40

AI Summary: Eric Schmidt, former Google CEO, offered insights on distinguishing founders from executives, explaining startups as a "validation game" for acquisitions, highlighting the resilience needed in founders when facing competition. Schmidt firmly believes we are not in an AI bubble, sees AI as undervalued, and expresses significant concern about China's rapid advancements in AI development and the potential negative global effects if the US doesn't accelerate its efforts, highlighting the tension between open and closed-source AI strategies and the broader implications for global values.

In a recent conversation on the AI Founder Journey podcast, former Google CEO Eric Schmidt shared invaluable insights about leadership, startups, and the future of AI. Drawing from his vast experience as a tech executive, Schmidt offered a unique perspective on what makes startups successful, how to identify exceptional talent, and why the AI revolution might be more significant than we realize.



Schmidt begins by making an important distinction between founders and executives. While founders typically start with a passionate idea they develop at home, executives like Schmidt are professional managers who scale and grow businesses.

I don't consider myself a founder. Founders sit at home and they have an idea and they become very passionate about it. I'm much more of a professional executive. I work for you or with you to scale and grow the business.

This distinction played out perfectly at Google, where Schmidt joined Larry Page and Sergey Brin. Schmidt acknowledges their brilliance while recognizing his own contribution was in scaling the company. He recalls Page telling him, "You know, we don't need you now, but we're going to need you in the future," a statement Schmidt found initially obnoxious but ultimately accurate.


According to Schmidt, most startup ideas are features rather than full platforms or products. Most founders discover a better way of doing something that's actually an important component of a bigger system.

The tech ecosystem works through what Schmidt calls "a validation game." Multiple startups compete, and whoever emerges as the winner gets acquired by larger companies. As Schmidt puts it: "The startup founders are fighting essentially a validation game and whoever emerges among the startups is the one you should buy."

This system benefits large companies because they're not just paying for talent but also confirmation that the team is the best at what they do right now. Schmidt believes this model works exceptionally well and shouldn't be tinkered with, given how it has created enormous wealth and technological advancement through companies like Microsoft, Apple, Google, Meta, Airbnb, Tesla, and SpaceX.


What does Schmidt look for when investing in founders? His answer reveals a fascinating insight into leadership psychology:

At some point, you're going to be in the situation where a big company has decided to do the same thing as your little company is doing... The founder I'm looking for looks at that and says 'game on' and then doubles their energy, redoubles their focus on scaling, redoubles and motivates their team against this competitive threat.

Schmidt explains that when no competitors exist, startups can exist in a comfortable bubble. The true test comes when competition arrives – that's when leadership qualities truly matter.


Schmidt believes CEOs are generally underappreciated. The role involves constant pressure from all directions: internal issues, hiring challenges, manipulative competitors, negative press, demanding customers, and impatient investors.

"It's so easy for us to criticize and we love to criticize. Everything's wrong. Everything's 'I could do so much better.' You try it. You take on those things," Schmidt challenges. "Do you have the fortitude to get up every day and for 12 or 14 hours a day deal with all of this and then go home and be nice to your family and don't kick the dog? It's very hard. It's much harder than people think."


With so many AI startups emerging, how should engineers decide which to join? Schmidt's advice might surprise you – don't try to pick the winner. Instead, "pick the place where you're going to have the most fun because if you're having fun, assuming you're a high-quality person, you're working with very smart people."

He reminds us that while we celebrate individual icons like Steve Jobs, most innovations come from teams of brilliant people working together. Young professionals should find the smartest people they know and work with them, taking risks while they can afford to do so.


For today's startups, Schmidt emphasizes one critical factor: "Show me where the learning is in your model."

He explains that the fastest path to wealth for a startup founder is building a company that incorporates machine learning – whether through foundation models, specialized models, or fine-tuned models. As learning accelerates, market share accelerates, potentially leading to market dominance.

If you're building a company that's not learning as it goes, and by that I mean using a foundation model or some other AI tool, you're going to be beaten by the competitor that has learned that because the fastest learner wins.

The ideal AI company, according to Schmidt, combines AI researchers who understand machine learning with engineers who can build systems. When you achieve learning at scale, your market share can accelerate toward 90% dominance.


When recruiting top talent, Schmidt advises creating "a form of seduction" by presenting an incredible opportunity paired with a hard problem. The best people aren't primarily motivated by money or fame – they want to solve challenging, fundamental problems.

What I've learned about people is if you can motivate them around a mission, it's so much easier to manage them because they're busy. I don't need to worry about them. I know they're busy because they actually care.

Schmidt recalls potential hires at Google who were fixated on specific titles rather than solving hard problems. Those who insisted on particular roles over tackling challenging work missed out on significant opportunities, including potential fortunes.


Schmidt makes an important distinction between two types of challenging personalities: divas and naves.

A diva is Steve Jobs. Brilliant, irascible, committed. No one ever questions Steve's commitment to his customers and to Apple and to the success of his firm. A nave is someone who is in a similar position, but they're fundamentally self-dealing. They're just optimizing for themselves.

His advice? Keep the divas – they're the ones who drive innovation and change companies. Support them because they deliver new products, but you have to listen to them. The naves, however, need to be fired because their self-interest will never align with the company's mission.


When asked if we're in an AI bubble, Schmidt responds with a definitive no. In fact, he believes AI is "undervalued, underappreciated, not overappreciated." He explains that the AI industry is governed by scaling laws:

• The first scaling law involves deep learning, which gave us ChatGPT, Gemini, and Claude 3
• The second involves reinforcement learning (planning), exemplified by OpenAI o3 and DeepSeek R1
• The third is "test time compute," where models learn as they go

None of these scaling trends have shown signs of slowing. While we may eventually hit physical limitations like electricity constraints, hardware improvements continue with impressive new chips from Nvidia, AMD, Google, and others. Schmidt sees tremendous room for continued growth in AI's ability to help with business processes, reasoning, invention, and scientific discoveries.


Reflecting on a world with artificial general intelligence (AGI), Schmidt references his book "Genesis," co-authored with Henry Kissinger and Craig Mundy, which explores the implications of superhuman intelligence.


Schmidt poses a thought experiment: "Let's imagine that you and I are in charge of a system which is smarter than all humans... How would you manage it? What would you allow it to work on?"

He raises crucial questions about preventing harmful uses, preserving human freedom, and handling scenarios where AI optimizations conflict with human needs or emergencies. The challenge, Schmidt argues, is that "we don't have a system of thinking in our current language about what it's like to live with such intelligences."

However, he also emphasizes the extraordinary potential: properly applied AGI could solve climate change, extend human life, and address diseases that have troubled humanity for centuries.


Schmidt expresses serious concern about China's progress in AI, particularly following the success of DeepSeek:

I want to be very clear here. China is going to win this race with enormously negative effects unless we get our act together.

He explains that China is pouring billions into AI development, and despite chip restrictions, they're developing impressive new algorithms. This creates a concerning dynamic where China appears to be pursuing an open-source strategy in response to America's closed-source leadership. The risk is that countries without access to leading closed-source models will adopt Chinese open-source alternatives, potentially spreading Chinese influence.

Schmidt emphasizes the importance of ensuring that AI develops with American values of freedom of speech, expression, and movement, rather than under a Chinese model that might neglect these principles.


Schmidt describes the tension in AI development between open and closed approaches. While he grew up in the open-source movement and recognizes its value in attracting talent, he acknowledges the economic reality that companies spending $100-200 million on model development are unlikely to simply give away their work.

He advocates for more open-source innovation in universities to generate original ideas while accepting that top commercial models will likely remain proprietary. The challenge is balancing innovation with safety concerns as models become more powerful, potentially developing "dangerous knowledge" as they reach deeper levels of understanding.


Eric Schmidt's insights reveal a tech landscape poised for tremendous change. His experience scaling one of the world's most influential companies gives weight to his predictions about AI's transformative potential and the competitive dynamics that will shape its development.

For founders, executives, and engineers navigating this rapidly evolving field, Schmidt's advice emphasizes fundamentals that transcend technological shifts: focus on solving hard problems, build learning systems, recruit mission-driven talent, and maintain the courage to compete even against formidable opponents.

As we stand at what may be the beginning of the most significant technological revolution in human history, Schmidt's perspective reminds us that while the tools and capabilities may change dramatically, the human elements of leadership, innovation, and competition remain essential drivers of progress.