Google CEO's speech at Stanford University
It's quite amusing. Eric Schmidt, who served as Google's CEO for 10 years, was invited to a conference at Stanford University's School of Computer Science a few days ago. During his speech, he was completely unfiltered. At one point, he earnestly told the students in the audience that the conference was confidential and they should not leak anything he said.
However, the organizers informed Eric Schmidt that the conference was being live-streamed... His expression froze. Although Stanford later removed the video from YouTube, many people had already archived it, and there’s a full transcript backup on Github: transcripts
/Stanford_ECON295⧸CS323_I_2024_I_The_Age_of_AI,_Eric_Schmidt.txt
As usual, here are the key takeaways:
- Why is Google losing so badly in the AI field now? Because Google believes that letting employees go home early and maintaining work-life balance is more important than winning the competition. If your employees only come to the office one day a week, how can you possibly compete with OpenAI or Anthropic?
- Look at Elon Musk, look at TSMC. The reason these companies succeed is because they can push their employees hard. You have to pressure your employees enough to win. TSMC makes physics Ph.D.s work in factories during their first year. Can you imagine American Ph.D. students working on assembly lines?
- He admitted to making many mistakes, such as once thinking NVIDIA's CUDA was a stupid programming language. But now, CUDA is NVIDIA's strongest moat. All large models run on CUDA, and only NVIDIA's GPUs support CUDA. This combination is unshakable by other chips.
- He also found it unbelievable when Microsoft partnered with OpenAI. How could Microsoft outsource its most important AI business to such a small company? Once again, he misjudged the situation. And look at Apple's lukewarm approach to AI—big companies have truly become bureaucratic. The real hustlers are in startups.
- TikTok taught Americans a lesson. If any of you young people start a business in the future, steal music or whatever you need to do—seemingly mocking TikTok’s early tolerance of pirated BMG—if you succeed, you’ll have money to hire top lawyers to clean up the mess. If you fail, no one will sue you.
- OpenAI’s Stargate was initially promoted as needing $100 billion, but in reality, it might cost over $300 billion. The energy gap is too huge. He suggested to the White House that the U.S. should either build strong ties with Canada—rich in hydropower, cheap labor, and close by—or cozy up to Arab countries for sovereign investment.
- Europe is already hopeless. Brussels (where the EU is headquartered) has always been destroying opportunities for tech innovation. Maybe France has some hope, but Germany is a lost cause, and the rest of Europe is even worse. India is the most important swing state among U.S. allies, and the U.S. has already lost China.
- Open source is great, and most of Google’s historical infrastructure benefited from it. But honestly, the costs in the AI industry are too high for open source to bear. The French large model Mistral, which he invested in, will switch to a closed-source model. Not all companies are willing or able to be a sucker like Meta.
- AI will make the rich richer and the poor poorer, and the same goes for countries. This is a game for powerful nations. Countries without technical resources need to secure a spot in the supply chain of powerful nations, or they’ll miss the feast.
- AI chips belong to high-end manufacturing, with high output value but unlikely to boost employment. Few of you have probably been to a chip factory—it’s all mechanized production, no humans needed. Humans are dumb and dirty. So don’t count on a manufacturing revival. Apple moved MacBook production back to Texas not because wages are lower there, but because they don’t need to hire many people anymore.
- Historically, electricity didn’t create more productivity than steam engines when first introduced to factories. It took about 30 years for distributed power to redesign workshop layouts and enable assembly systems, leading to a productivity leap. AI today is like electricity back then—valuable, but it still needs organizational innovation to deliver massive returns. Right now, everyone is just picking the "low-hanging fruit."
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.