China’s AI Is Everywhere. But Where Is the Money?

 

I came across a report from Rhodium Group about China’s AI industry, and one number immediately caught my attention.

Chinese AI models are reportedly generating only around 10% of the revenue generated by OpenAI and Anthropic.

That sounds surprising when you think about how much we've been hearing about China's AI progress.

DeepSeek, Alibaba, ByteDance, Z.AI, Moonshot, MiniMax… the names keep popping up whenever there's a conversation about the global AI race.

But when you look at the money, the picture suddenly looks very different.

Rhodium estimates OpenAI at around $40 billion in annual recurring revenue, while Anthropic is estimated at around $65 billion. Chinese AI companies are nowhere close collectively. ByteDance is estimated at around $4 billion, Alibaba at $2.4 billion, while companies like DeepSeek are much smaller by comparison.

And this is where I think things get interesting.

Being popular doesn't automatically mean being profitable

We often use downloads, users, benchmark scores and how quickly a model spreads as proof that an AI company is winning.

But eventually, someone has to pay the electricity bill.

You can have millions of people using your AI model and developers building products around it. But if you can't turn that usage into sustainable revenue, you have a very different problem.

China seems to be facing exactly that challenge.

A lot of Chinese AI models are open source. That's great for adoption because developers can use the models and build around them.

But from a business perspective, there's an obvious question:

If everyone can use your model without paying you directly, how exactly do you make money?

That's much harder.

Then there's DeepSeek

DeepSeek is probably the perfect example of why the AI race isn't simply about who has the best model.

When DeepSeek became globally famous, it felt like a huge moment. Suddenly everyone was talking about cheaper AI models and questioning whether the enormous spending by American AI companies was really necessary.

I still think DeepSeek was an important moment.

But this report adds another layer.

Rhodium estimates DeepSeek's annual recurring revenue at only around $500 million, despite its enormous attention and valuation.

That makes me wonder whether investors are sometimes paying for the future of AI rather than the AI business that exists today.

Those are two very different things.

China isn't sitting still

I wouldn't read this report as “China has lost the AI race.” That's way too simple.

Rhodium estimates Chinese AI infrastructure spending could reach around 932 billion yuan ($139 billion) in 2026, roughly double the previous year.

So China is clearly investing heavily.

The interesting part is that spending and revenue aren't growing at the same level.

And that could become one of the biggest battles in AI over the next few years.

Building an impressive AI model is one thing.

Building an AI company that can continuously pay for GPUs, data centers, researchers, engineers and electricity is something else entirely.

Maybe we're measuring the wrong scoreboard

For years, we've talked about the AI race almost like it's the Olympics.

Who has the smartest model?

Who scores highest on benchmarks?

Who has the biggest model?

But maybe the more important question is:

Who can turn AI into a sustainable business?

Eventually, AI companies need customers and recurring revenue. They need enough cash to keep buying compute and developing better models.

And they need a business model that still works when investors stop throwing billions at them.

Interestingly, Chinese AI companies are still looking for ways to monetize their models more effectively, while facing financing challenges.

My takeaway

After reading the report, I don't think the interesting story is that China is “losing” or America is “winning.”

The more interesting story is that AI is entering the boring part of the race.

The flashy demos are here. The insane benchmarks are here. The viral models are here.

Now comes the annoying adult question:

“Okay… but how are you actually going to make money?”

And honestly, I think that question is going to become much more important over the next few years.

The AI race might not ultimately be won by whoever builds the coolest model.

It could be won by whoever figures out how to turn intelligence into a sustainable business without burning through an ocean of money.

And honestly, that's a much more interesting AI race to watch.

Source : YahooFinance Link