Anthropic, the parent company of Cloud, a top global large-model company, has successfully gone public.

This time, it raised more than $80 billion, making it the largest IPO in history.

After going public, Anthropic’s market value briefly exceeded $2 trillion, making it the world’s sixth-largest company. Although it closed at just $1.27 trillion, it still ranks as the world’s twelfth-largest company by market capitalization.

Given Anthropic’s impressive performance, I imagine the most disappointed are Baidu and Robin Li.

Why do I say that?

Many people may not know that more than a decade ago, Anthropic founder Dario Amodei worked at Baidu, where he was primarily involved in AI research and development.

Baidu established its Institute of Deep Learning (IDL) as early as July 2013, with Robin Li personally serving as its director and Kai Yu as executive deputy director. This also signaled Baidu’s determination to enter the AI field.

In 2014, Baidu established an AI lab in Silicon Valley, the United States (SVAIL). The lab was led by renowned researcher Andrew Ng, who also recruited research staff in the U.S.

At the time, Dario Amodei was pursuing postdoctoral medical research at Stanford University School of Medicine—a job that had virtually nothing to do with AI.

However, he had majored in physics as an undergraduate and had strong programming skills. Some leaders at Baidu’s AI Lab happened to come across code he had written.

Baidu then recruited him to its AI research team, where he mainly worked on a speech recognition AI project. This was Dario Amodei’s first encounter with AI and the beginning of his education in the field, laying a solid foundation for his later career in AI.

During his time at Baidu’s AI Lab, he learned a crucial principle: the “scaling laws.”

While conducting speech research at Baidu, Dario Amodei realized that the more data and computing power you give a model, and the more thoroughly you train it, the better it performs. This empirical formula has since become a fundamental principle behind the development of today’s large language models.

At the time, among the major tech companies around the world, Baidu was the only one truly capable of helping him discover this principle, because it had vast amounts of training data—a resource unmatched anywhere else in the world.

That’s why his experience working at Baidu’s AI Lab had such a profound impact on his later career.

However, Dario Amodei left Baidu after working at its AI Lab for just one year. He has never publicly explained why he left.

After leaving, he worked briefly at Google, then joined the AI startup OpenAI in 2016.

What he learned at Baidu’s AI Lab helped him make a big impact at OpenAI. He played a leading role in developing GPT-2 and GPT-3. OpenAI later grew into one of the world’s leading AI model companies, and Dario Amodei became its vice president.

In 2022, however, he left with several former OpenAI colleagues and founded Anthropic. In just a few years, they turned Claude into one of the world’s leading large language models.

Dario Amodei’s career shows just how formidable he is. He has not only exceptional learning ability but is also highly capable of leading teams.

Seeing what Dario Amodei has achieved today, I wonder what Baidu and Robin Li must be thinking.

Remember, in 2014 Dario Amodei was just an ordinary researcher at Baidu, while Baidu was already one of China’s three biggest internet companies. That year, Baidu’s revenue exceeded 49 billion yuan and its net profit topped 13 billion yuan.

By 2025, Baidu’s revenue had grown to around 130 billion yuan, but its net profit attributable to shareholders of the parent company was just 5.6 billion yuan.

According to financial data released by Anthropic, its revenue reached $33.24 billion in the first three quarters of 2026, equivalent to about 220 billion yuan, an increase of approximately 1,200% over the same period last year.

Additionally, as of October 2026, its annualized revenue had exceeded $81 billion, equivalent to more than 540 billion yuan.

Based on this, Anthropic’s revenue could exceed $100 billion next year, equivalent to more than 670 billion yuan—roughly four times Baidu’s revenue or more.

Bear in mind, this company was founded just four years ago, while Baidu has been around for 26 years.

The reason Anthropic has been able to surpass the established internet giant Baidu in such a short time is its powerful AI.

The irony is that Baidu was where Anthropic’s founder first learned about AI; it was Baidu that set him on this path. Without Baidu, perhaps Dario Amodei would still be doing medical research today instead of becoming one of the world’s AI giants.

This achievement should have belonged to Baidu—not merely because Dario Amodei once worked there, but because Baidu was one of the earliest companies in the world to research AI.

More than a decade ago, Baidu was one of the world’s leading AI companies, especially in speech recognition, where its AI was far ahead of the competition.

When Dario Amodei was working at Baidu’s AI Lab, he worked on Deep Speech 2, a speech recognition project.

When Deep Speech 2 was released, it was truly far ahead of the competition. Its recognition accuracy outperformed competing products from Google, Apple, Microsoft, Facebook, and others by more than 10 percentage points. Its word error rate for phrase recognition was as low as 3.7%, and the project was named one of MIT Technology Review’s Top 10 Breakthrough Technologies of 2016.

Thanks to these achievements, Baidu was generally ranked among the top three AI companies in the world at the time.

But times have changed. More than a decade later, Baidu is nowhere to be seen among the top ranks of the world’s leading AI companies.

Baidu is still working on AI, and AI has become one of the main drivers of its business growth. It has also achieved some success in areas such as large language models.

But there’s no denying that, as of today, Baidu’s AI has gone from first-rate to second-rate.

You have to believe that Baidu has always had remarkably foresightful vision. Many of the businesses it pursued came much earlier than its competitors did.

Take AI, for example: Baidu was also at the forefront globally, and at one point even declared “All in AI.”

But with AI booming around the world today, why has Baidu been reduced to a mere bystander? Why does Baidu always get an early start but arrive late? What exactly is the problem?

I think there are two key reasons.

The first is that its business is too diversified.

A close look at the stronger AI companies shows that they are highly focused. They concentrate mainly on AI and related businesses, rather than dabbling in all sorts of unrelated areas. This lets them put their money where it matters most.

Baidu, by contrast, has had a fairly scattered business portfolio over the past decade or so. Besides its traditional search business, it has tried many other things. Apart from large language models, cloud services, and autonomous driving, which are related to AI, quite a few of its ventures have had virtually nothing to do with AI.

For example, it ventured into food delivery and even lending—neither of which had anything to do with its core business or AI.

And even in AI, it hasn’t been all that focused. In large language models, for instance, it does everything: ERNIE Bot, ERNIE Code, Miaoda, digital employees, AI ecosystem marketing, and so on. It wants to do everything, but doesn’t do any of it well.

If Baidu could narrow its focus and devote more resources to one area, such as large language models or chips, perhaps it would do better.

The second reason is a lack of execution.

Competition in AI is now fierce, and the pace of change is extremely fast. To gain a foothold in such a cutthroat market, a company needs strong execution, as well as a very stable team.

But judging by how Baidu’s various businesses have been implemented, its execution doesn’t seem particularly strong. Many projects start with a bang and then fizzle out: the initial vision looks promising, but as implementation progresses, they gradually seem to disappear.

I think this weak execution is closely related to the instability of Baidu’s own teams. In particular, over the past few years, leaders have frequently changed across several business lines. Large language models, autonomous driving, and the mobile ecosystem have all seen leadership changes.

There have also been frequent internal reorganizations, including changes to business groups, reporting lines, and departmental structures. This has significantly weakened the teams’ effectiveness and continuity.

So even though Baidu entered many businesses earlier than its competitors, when it comes to actually launching products, it always seems to be a step behind. #Anthropic上市次日大跌,股价接近破发 $BNB

BNB
BNB
524.96
-7.92%