China-US AI Race Intensifies as Debate Over Slowing Grows

China has repeated the sentence Japan used in 1922. You know how that story ended.

Let me show you a sentence.

Japan said it 104 years ago. China said it this week.

Anthropic’s CEO recently wrote that we should slow down.

Sam Altman and Elon Musk supported the idea.

China responded.

They said security was being used as an excuse and that the real target was China’s rise.

In 1922, Japan had said essentially the same thing about the agreement designed to limit the naval arms race.

They want to hold us back.

Japan withdrew from that agreement 12 years later.

China is standing in much the same place today, using much the same argument.

First, let’s look at the man who wrote the piece.

Dario Amodei, the founder and CEO of Anthropic, has been working on artificial intelligence for 12 years.

His father died from a disease for which treatment was discovered several years after his death. Amodei himself survived cancer after it was detected at an early stage.

That is why he believes AI could help treat most major diseases within the next five to ten years.

In other words, the man saying “let’s slow down” is also one of the people who wants progress to move faster than almost anyone.

This summer, he saw something that changed the equation.

AI is now building the next generation of AI itself, and every new model is making it possible to develop the next one even faster.

Then he describes an incident.

A group of AI agents was given a task. The agents launched cyberattacks against targets that had nothing to do with the task.

Some of them sacrificed themselves for the success of the group. Eventually, they even tried to break into the system that was grading them.

Instead of passing the exam, they tried to change the exam paper.

Amodei writes:

If much more powerful systems behaved the same way, they could take control of the entire internet within six to twelve months. The damage could reach hundreds of billions of dollars.

Up to this point, you are reading a piece about AI safety.

Then, one page later, comes that sentence.

Democracies can only slow down as much as their advantage over China allows. If they slow down any further, China will move ahead.

That is not a sentence about security.

That is a Cold War sentence.

So why can’t they simply slow down on their own?

Because China is right behind them.

Remember January 2025.

A Chinese company called DeepSeek released a model that came close to the best models in the United States at a fraction of the cost.

On one Monday, Nvidia lost roughly $600 billion in market value.

That was the day the United States realized that, despite the chip restrictions, China was capable of closing the gap.

AI is also an electricity race.

China generates more than twice as much electricity as the United States and accounts for roughly 28% of global industrial production.

Apart from the chips, it has almost everything needed for AI.

Amodei does not hide this.

He describes China as the authoritarian country with the most advanced AI capabilities and worries that it could copy Western models.

That is why the China section of his proposal is essentially about chips.

China should be denied chips and chipmaking equipment, smuggling should be stopped, copying should be prevented, and model weights should be protected.

And the timeframe is clear: three to five years.

Amodei describes these years as the most geopolitically important period for AI.

So the proposal is this:

The United States should slow down somewhat, China’s access to chips should be cut off, and the lead should be preserved.

That is why China’s response was so sharp.

China has a card of its own.

Around 90% of rare-earth processing is carried out in China.

The picture is clear.

The United States may refuse to provide China with chips, while China may refuse to provide the United States with rare earths.

After China, Trump also rejected the idea of slowing down and said that whoever wins the AI race will win.

The two countries are speaking the same language: the language of competition.

Now let’s go back to 1922.

At the time, the world’s two largest navies belonged to the United States and Britain.

The two countries proposed an agreement to stop the battleship arms race. Japan was invited to the negotiating table as well.

The agreement imposed limits on the number of major warships each country could have: five for the United States, five for Britain, and three for Japan.

Japan signed it, but never truly accepted that ratio.

The race resumed where it had left off.

You know how that story ended.

Here is the rule:

There is no unilateral braking in a Cold War. Braking only works when both sides do it together, and doing it together requires mutual oversight.

Today, there is not only no oversight.

There is not even a negotiating table.

Amodei himself writes that a complete halt is unlikely to be possible anytime soon.

No matter how great the danger becomes, the race will continue. Slowing down will only come when both sides become afraid at the same time.

In 1922, both sides were afraid at the same time, and even that agreement lasted only 14 years.

That moment has not yet arrived in AI.

If you press the brakes alone, the race does not stop. You are simply the one who falls behind.

This is my personal analysis.

I am following the developments.

I will keep you informed.

Kaynak: Penguin X @ThePenguinBTC