news

The truth has always existed, but it has never been spoken out loud by some high-profile figures.

Therefore, when these truths that have long been a consensus among people are accidentally spoken out, they will trigger discussion.

The latest to go wrong was former Google CEO Eric Schmidt, who was a guest speaker at a Stanford class and completely let himself go.

The 69-year-old man, with a ruddy face, spoke earnestly to the students in the audience and told them: First, keep it confidential and don't spread it; second, don't learn it. Until the organizer told him: this meeting was live broadcast by cameras...

As luck would have it, Schmidt revealed many truths in the current AI circle during the nearly one-hour Q&A session.

For example, open source means not making money, and open source means contribution.Although open source is correct, few are like Meta. The third version of Mistral, which he invested in, is likely to be closed source because the training cost is too expensive and they need revenue.

For example, implementationAGIWe still have a long way to go, and we don’t have enough money and energy.When promoting OpenAI's Stargate, Sam Altman said it would cost $100 billion. In fact, $300 billion is not enough. A large amount of money was used to purchase hydropower resources, which is exactly where the United States is "stuck".

Of course, as the former CEO of Google for 10 years (2001-2011) and a long-time employee of the US Department of Defense, he also rarely reveals some of his true thoughts:

As for Google's current backwardness, he believes that it is because its employees are too idle (this sounds familiar), unlike TSMC and Musk's companies that are so aggressive;

In the business world, he told students directly, there is no such thing as data privacy.According to past experience, if you are successful enough, you can pay for the best lawyer to win the case for you, and if you are unsuccessful, no one will sue you;

He also commented on the world's leading countries in artificial intelligence one by one: the EU is not good enough, Germany is not good enough, China is good enough but China is a rival, France and India can be won over.

These remarks immediately sparked controversy, so much so that the Stanford account that posted the video urgently removed the original video from YouTube.

But the full text of this video and conversation has long been backed up by everyone, and people continue to spread it on social media. These big truths are being discussed by more people, and it also allows people to get a rare glimpse into the true thoughts of this pretending to be confused "Mr. Shi" as he travels between Silicon Valley and Washington.

Let’s first take a look at the “truth” about the industry that he accidentally revealed.

At the beginning, he complained about Nvidia for everyone. "Why is Nvidia worth $2 trillion while other companies are still struggling?" Schmidt asked the students present.

In fact, the entire AI industry chain has long suffered from Nvidia, whether it is a startup or a giant.OpenAICompanies of this level have to act according to Nvidia's wishes.

"If you have unlimited money, anyone would choose Nvidia B200 because it's faster," Schmidt said. "I talked to AMD's Lisa for a long time, and although they have built Rokam to completely convert CUDA, no one is using it."

But Schmidt also expressed his helplessness towards this monopoly: "Is there anyone here whose computing device does not have an Intel chip?" Unfortunately, no one raised their hand. The foundation laid by Intel in the 1990s still dominates the market today. It can be inferred that Nvidia will still dominate the development of the entire AI trajectory for a long time to come.

At the same time, the entire AI will evolve towards a more polarized pattern."Capital-intensive giants will spend huge sums of money to build data centers, and software vendors will also switch from open source to closed source to further consolidate the giants' monopoly position."

This almost shatters the beautiful fantasy of AI universalization and open source benefiting application development. Many open source big models emphasize this core value. "Open source is good, and most of my career and Google's foundation are built on open source. But big models are different. They are too expensive, and no one can keep being ripped off like Meta."

This is not only true for the entire business landscape, but also for the national level. The mastery of artificial intelligence technology is also a game for rich countries, requiring a lot of capital, technical talent and strong government support. The rich will get richer and the poor will stay poorer.

Schmidt also made no secret of the fact that "China has risen rapidly, and the United States needs to increase investment to maintain its dominant position. It was the US government that banned the export of Nvidia chips to China, and this advantage allowed the United States to lead by several years."

Of course, he also sharply commented on the countries currently active in the field of artificial intelligence in the world: "In my opinion, India is a wavering country. Japan and South Korea are in our camp, but they are useless. There are no good options elsewhere (excluding China in the context). The EU has messed up because of the restrictions of various Brussels bills. France is struggling for this, but Germany will not."

Another "heartbreaking" truth that no one wants to admit is: we are still a long way from AGI.

No matter how the market atmosphere is or how excited the public sentiment is, it can be seen from the fact that OpenAI has been slow to release GPT-5 in the first half of the year that the technological progress of the entire basic large model is stagnant.

“I’m talking to some big companies and they’re telling me they need $10 billion, $20 billion, $50 billion, or even $100 billion. OpenAI said Stargate would take $100 billion, which was difficult, but Sam Altman thinks it might take $300 billion.”

And money alone is not enough. "If you want to build a $100 billion to $300 billion data center, electricity will become a super scarce resource."

To this end, he even went to the White House in person to "meet" and said that the United States needs to build an alliance from now on: "I told them that we need to be best friends with Canada. They have a lot of hydroelectric resources that our country does not have; another option is to let Arab countries fund us. They have money."

In addition to the lack of money and energy, the industry itself is far from mature. Schmidt compared AI to electricity. Referring to past history, in the early days of electricity development, it did not instantly replace the productivity created by steam engines. Instead, it took about 30 years for the structure of the entire society to change, and distributed power sources transformed the layout of workshops, and then electricity began to achieve a leap in productivity.

This is not something that can be driven by a single company, but requires the synchronous evolution of the entire society. Schmidt emphasized that he is looking forward to a new way of software programming that may more than double the productivity of the current way.

Schmidt was not the only one who knew these "truths". The Verge's report described that these truths had long been a private consensus among these executives, but this time Schmidt spoke them out loud in an accident.

These truths might not cause much controversy in themselves, but after the video was released, Silicon Valley's opinions and discussions about him almost all revolved around one thing: his attack on Google's work-from-home policy.

When he was asked why Google lost its dominance in AI and was overtaken by OpenAI and Anthropic, he said bluntly:

“I’m no longer a Google employee, but apparently Google decided that work-life balance, going home early, and telecommuting were more important than winning the competition. Startups succeed because their people work their ass off. I’m sorry to be so blunt, but that’s the truth. If you’re starting a company right out of college, you can’t allow people to only work from home or come in one day a week if you want to compete with other startups.”

In other words, he believes that one of the reasons why Google falls behind is that it is not "competent enough" and does not push its employees too hard.

He even said, "I don't like Musk's personal behavior." But he praised Musk for working day and night. "Once we chatted, we were in Montana, and he was going to fly that night to attend a midnight meeting with x.ai."

He also affirmed TSMC's spirit of sending PhD students to the grassroots: "I was deeply impressed by a rule at TSMC that newly graduated PhD students, even excellent physicists, must work in the factory's basement in their first year. Can you imagine an American physics PhD doing this?"

Not only did they scold the workers for not working hard enough, they also scolded them for not being diligent enough.

Google employees were the first to launch a counterattack, and all kinds of insults flooded into social platforms.

for example,DeepMindDirector of Research at GoogleSora" Dumitru Erhan, head of Veo, directly scolded:

Complete rubbish from every angle.

Some non-Google employees were also dissatisfied with Schmidt's violation of the "work-life balance" that Silicon Valley is most proud of. Some people teased him about Nvidia:

I know a company that has always been leading, but it also doesn't force its employees to come to work.

Because the opinions were so strong, Schmidt even said in a report in the Wall Street Journal that he would retract his criticism of working from home.

This is something I have never seen before. It turns out that true words can be taken back.

This has made the topic of "not being competitive enough" the focus of discussion. However, if you really watch his entire sharing, you will find that there are more important topics than this. In particular, for the Chinese context, discussing 996 with Silicon Valley is not the same thing, and in this heart-to-heart talk, Schmidt actually said a lot of his true views on China and Chinese companies.

You know, today Schmidt basically has two identities. First, he is still a major shareholder of Google. According to reports, he still holds 140 million Alphabet shares, worth $24 billion. Google's backwardness will still affect his income. As a former leader, he chose to throw all the problems on his employees, even though almost everyone today points to Google's complicated bureaucracy when discussing Google's backwardness. Doesn't he know? He knows, but he has to pretend to be confused. Is he criticizing Pichai by name?

The other identity is even more important. Schmidt is no longer an "entrepreneur" but aHe is a broker active in American politics, and is the most typical person who advocates the theory of China's technological threat, directly participates in the US's sanctions and suppression strategy against China's technology industry, and profits from it.

Public information shows that as early as during the 2008 presidential election, Schmidt joined Obama's campaign team and assisted Obama's campaign by providing technical support. In March 2016, Schmidt served as chairman of the Department of Defense Innovation Advisory Board and continued to provide advice to the Pentagon until his resignation in November 2020. The US media reported that "he jumped from a Wall Street tech tycoon to a very influential figure in Washington, and used the "revolving door" mechanism between politics and business to "jump left and right" between the two identities."

Since 2019, he has repeatedly mentioned in public the threat posed by China's technological progress to the United States, from chips to5GTo AI, from Huawei to TikTok to today's big model companies in China.Recently, he even said directly in a conversation: "China's AI development basically relies on the open source of American models."

On the other hand, according to the compilation of many American media and institutions, Schmidt has invested in many AI companies through his own investment company. These companies are often related to the American interests he advocates and have received large orders from the US government.

Microsoft's former Chief Technology Officer (CTO) Myhrvold once pointed out this problem: "When you gain support by intimidating the public through a rhetoric of bellicosity, fear, and conflict, you run into a real problem - you may end up shooting yourself in the foot."

This time in the Truth or Dare game at Stanford, Schmidt also gave his true thoughts on his own actions.

"In order to make artificial intelligence more useful, I have become an arms dealer. Yes, it's legal," he said.

Schmidt also served as chairman of the U.S. government's artificial intelligence commission, which led a series of crackdowns on China.

"I chaired an AI commission that looked at this very carefully. You can go read the report, it's about 752 pages. Let me summarize it briefly. We are currently leading, we need to stay ahead, and we need a lot of funding to do it. Our clients are the Senate and the House of Representatives, and out of that came a series of policies such as the CHIPS Act," he said.

“If the frontier models continue to advance, and if there are a few open source models, there may be only a few countries that can play this game. Sorry, which countries do I mean? Those with a lot of money and talent, strong education systems, and the will to win. The United States is one of them, and China is another.”

He also directly stated the real purpose of the US chip ban on China:

"So the US government banned essentially the NVIDIA chips, although they weren't allowed to say that was what they were doing, but they actually did that into China."

Another topic is TikTok, which Schmidt mentioned many times in his sharing that day.

"TikTok is not really a social media platform, but more like a form of television. The average TikTok user in the United States uses the app for 90 minutes a day and makes 200 videos, which is a lot of usage," he said.

“In the upcoming global elections, most misinformation will appear on social media, and social media companies are currently not organized enough to effectively regulate this information. For example, TikTok has been accused of favoring certain kinds of false information, although I have no evidence.”

He would not say that there was no evidence when he was on TV accusing Chinese technology companies. And he did not stop there, but continued to share his true feelings.

Although he has always claimed that China's technological progress is entirely due to copying the United States, he revealed his true thoughts in the conversation:

“The government is trying to ban TikTok, and we’ll see if that happens,” he told the Stanford students.If TikTok is banned, I suggest that each of you do something about your language model (LLM) issued the following instructions: Copy a TikTok,Get all the users and music, add my preferences to it, build this program in 30 seconds, release it in an hour, and if it doesn’t go viral, make some similar changes. That’s command. Do you see how powerful this is? If you can go from any language to any number of commands, that’s essentially what you have in Python.”

He further explained, but it only made things worse.

"I'm not telling you to illegally steal everyone's music. But if you become a Silicon Valley entrepreneur, which you all probably will be, you're going to do this: When this product takes off, you're going to hire a bunch of lawyers to clean up your mess, right? If no one uses this product, it doesn't matter if you stole it."（What you would do if you're a Silicon Valley entrepreneur, which hopefully all of you will be, is if it took off, then you'd hire a whole bunch of lawyers to go clean the mess up, right? But if nobody uses your product, it doesn't matter that you stole all the content.）

These true feelings revealed that these masters of playing dumb rarely revealed some of their real thoughts. However, these suggestions for future entrepreneurs are too different from their righteous public statements. The next time he sits in front of the TV and continues to sell the technology cold war and Chinese plagiarism, people will think of what Mr. Schmidt said in this truth or dare. Compared with discussing whether Silicon Valley should work 996, perhaps this is a more worthy topic of attention, and it may also be the reason why Schmidt wants to remove this video the most.