About China About DeepSeek Four major US technology giants warn the US government

Source: Xi Xiaoyao Technology Says

On May 9th, North American time, the U.S. Senate Commerce, Science, and Transportation Committee held an AI hearing. The theme of this parliamentary hearing was:

"Winning the AI ​​Race: Enhancing America's Capabilities in Computing and Innovation"

Translated, the core is what Americans are concerned about, how to keep the United States leading in the field of artificial intelligence.

Representatives of the four major American technology giants attending the hearing were:

• Sam Altman, co-founder and CEO of OpenAI

• Lisa Su, CEO and chairman of the board of directors of AMD

• Michael Intrator, co-founder and CEO of CoreWeave

• Brad Smith, Vice Chairman and President of Microsoft

CoreWeave, Inc. is a cloud computing company that leases GPU computing power. It completed its IPO in March this year and is currently valued at approximately $23 billion.

The entire hearing lasted more than three hours, covering infrastructure, talent training, energy demand, the social impact and risks of artificial intelligence, and many other aspects.

I have selected some of the "highlight moments" of the hearing, let's take a look.

II. Who will win the AI ​​competition between China and the United States?

The core question of this hearing is the following:

Let's see how the CEOs answered.

OpenAI CEO Sam Altman's answer directly admitted

AMD’s Lisa Su is similar -

The United States is currently leading in chips, but chip performance is not the only decisive factor. Even without using the most advanced chips, a lot of work can be done.

CoreWeave's Michael Intreto pointed out the current dilemma of the United States -

Although the United States is currently in the lead in AI infrastructure and the software stack for delivering infrastructure, under the current regulatory environment, it is difficult for American companies to obtain license approvals and the required electricity to continue to build large-scale projects.

Microsoft's Brad Smith pointed out more pointedly from a business perspective -

The key to this competition lies in whose technology can be most widely adopted by the rest of the world. The global market is determined by network effects, and 78% of the world's population lives outside the United States and China. At the same time, the examples of Huawei and 5G show that companies that enter the market first will be difficult to replace. The United States needs to export with appropriate controls and win the trust of the rest of the world.

The answers of the bigwigs were relatively implicit. They first said that the United States is currently in a leading position, but then changed the subject and immediately pointed out the current difficulties and risks from the field of their own companies.

3. Is DeepSeek a major threat?

This question directly hits the nail on the head!

See how Sam Altman answered -

Still pretending as always, "DeepSeek is not a particularly big deal. Of course, it does well in two areas: a good open source model and a good APP. Although the number of downloads exceeded ChatGPT in a short time, and they also have very good engineers who have made impressive results, OpenAI will also work hard to make better models."

Su Zifeng said that considering that DeepSeek was subject to various computing power restrictions, it was also a remarkable thing to come up with such a good open source model, but it was not a major breakthrough.

CoreWeave CEO said that DeepSeek made people realize that China is a strong competitor in the field of AI, reminding the United States that it must strengthen its competitiveness.

Microsoft President's view is similar to Su Zifeng. DeepSeek is in the middle of "significant progress" and "not so important", and specifically mentioned that "constraints encourage innovation to some extent" and "they hire college students who have no traditional experience."

Overall, they all directly or indirectly acknowledged DeepSeek's capabilities in the field of innovation.

4. Chip export control

There are already voices of opposition within the United States to the export control of AI technology and chips. The chairman of the meeting himself said:

Nvidia hopes to sell chips to the world.

At the same time, the CEOs were asked about their views on this issue:

“AI diffusion” refers to the export of AI-related software, hardware, and services from the United States to other parts of the world.

Altman's answer was still very direct.

AMD's Lisa Su implicitly expressed such a concern:

Although the US AI technology is very good, if foreign companies cannot use it due to export controls, they will choose solutions that can be used immediately, such as Chinese technology.

This is a very subtle way of opposing excessive export controls.

Microsoft President's views are very close to Su Zifeng's, and he expressed them more directly.

Eliminate the quantitative restrictions on these second-tier countries/regions that undermine confidence.

According to the definition of the US government, there are currently more than 140 second-tier countries/regions. The quantitative restrictions refer to the maximum number of 50,000 most advanced GPUs that these countries and regions can obtain between 2025 and 2027.

And he also expressed the following view in other parts of the hearing:

The key to winning the AI ​​competition between China and the United States does not lie in the quality of the technology, but whose technology can be most widely used globally.

He also cited the example of Huawei in the 5G competition:

Whoever seizes the initiative will be difficult to replace.

The implication is that the current export controls limit the competitiveness of American companies in other parts of the world.

And the CoreWeave CEO was even more direct:

I agree with Sam and Lisa Su. Moreover, he also complained a little about the previous export restriction policy that did not include his opinions, resulting in damage to the interests of their company:

V. Easter Egg: OpenAI to Release Open Source Model

The congressman asked a question about whether the United States has a leading position in open source models:

Regarding this point, Sam Altman gave a very direct answer:

Altman said that OpenAI will release a leading open source model this summer. In this case, let us wait and see.