Imagine this: a small Chinese startup builds an AI model as powerful as ChatGPT-4 — but does it for just $5.6 million. Now compare that to ChatGPT-4, which reportedly cost between $40–80 million. Crazy, right?
But it gets even more shocking…
The release of this budget-friendly AI model triggered a $600 billion stock market crash, and Nvidia took the biggest hit. Sounds unbelievable? Let’s break it down together.
Meet DeepSeek: The New AI Kid on the Block
In 2023, a Chinese startup called DeepSeek was founded by Liang Wenfeng. Just two years later, on January 10, 2025, they launched their newest AI model: DeepSeek R1.
What made headlines wasn't just its performance — but its price tag. It cost only $5.6 million to build, compared to the tens of millions companies like OpenAI spend on their models.
Why Did Nvidia Lose $600 Billion?
On January 27, 2025, Nvidia’s stock dropped 177%, wiping out nearly $600 billion in value. In fact, the entire U.S. stock market lost close to $1 trillion that day.
Why such a massive crash?
Well, Nvidia makes the powerful GPUs (graphics cards) that are used to train and run most AI models — including ChatGPT-4. If companies like DeepSeek can create similar models at a much lower cost, investors worry that Nvidia’s expensive GPUs won’t be in such high demand anymore.
In simple terms:
Less demand = less revenue = stock crash.
How Did DeepSeek Make AI So Cheap?
Training AI models is expensive — mainly because it needs a huge number of GPUs and a lot of electricity. But DeepSeek found a smarter way.
Instead of using tens of thousands of GPUs, like OpenAI does, they used a much smaller number. They also optimized how the AI model learns and runs, so it doesn’t need as much computing power.
As a result, DeepSeek’s AI model is up to 10 times cheaper to run than OpenAI’s models.
This makes powerful AI more affordable for smaller companies, not just tech giants. It opens the door for startups and mid-sized businesses around the world to join the AI race.
Free $44.99 Discount100% off 3 days left at this price!
It’s Open Source Too!
One big thing that sets DeepSeek apart? It’s open source.
That means anyone can download, modify, and use the AI model — as long as they have the hardware. In contrast, ChatGPT-4 is not open source, and its underlying code is protected.
DeepSeek is trying to break the monopoly in the AI world and make AI more accessible. More developers, more innovation — and possibly, a faster-moving AI industry.
But Wait… Isn’t Nvidia Still Involved?
Here’s an interesting twist: Even though DeepSeek claims to have built a cheaper AI, it’s still likely using Nvidia GPUs — just like everyone else.
In fact, DeepSeek is rumored to have used up to 50,000 Nvidia H100 GPUs, which are some of the most powerful (and expensive) chips available. Each H100 costs around $25,000 — not exactly pocket change.
But how did a Chinese company get so many GPUs when the U.S. banned exports of these chips to China?
How Did They Get Around U.S. Restrictions?
Since 2022, the U.S. has restricted the sale of advanced Nvidia chips like H100 and A100 to China, fearing they could be used for military or advanced AI development.
However, some Chinese companies have still managed to buy small amounts legally — often for educational or research use. But DeepSeek reportedly used thousands of these GPUs.
So how did they get them?
There are a few possibilities:
-
Leftover stock: Sometimes U.S. companies have extra chips they don’t use. These could end up being sold unofficially on the market.
-
Third-party imports: Chinese buyers might use companies in India, Taiwan, or Singapore to get around direct trade restrictions.
-
Underground markets: It’s possible some GPUs were obtained through unofficial channels or gray markets.
In any case, DeepSeek found a way — and that shows how determined and resourceful China’s tech scene can be.
Free $44.99 Discount100% off 3 days left at this price!
Using What You Have – And Making It Work
DeepSeek’s biggest achievement might not just be its AI model — but its ability to work with limited resources. Instead of waiting for cutting-edge chips, they optimized what they had.
This proves that you don’t always need the newest or best tech to build something powerful. With smart engineering, even older or less advanced chips can create impressive results.
China vs. U.S. – The Bigger AI Battle
Some people worry that using DeepSeek could mean China collecting your data. But let’s be real — the U.S. has been doing that for years too.
The real battle here isn’t just about AI models — it’s about data control and infrastructure. Who controls the technology, and who controls the information?
That’s where things start to feel less like a tech competition and more like a new kind of digital Cold War.
Final Thoughts
DeepSeek has shaken up the AI world. They showed that it’s possible to build high-quality AI at a much lower cost, and they opened it up to the world with an open-source model.
Meanwhile, companies like Nvidia are still essential because no matter who builds AI, they still need powerful chips — and Nvidia makes most of them.
So yes, AI is becoming cheaper, more accessible, and more competitive. But the demand for GPUs? That’s only going up.
Whether it’s the U.S., China, or another country leading the charge — one thing is clear:
AI is here to stay, and the race has only just begun.
Free $44.99 Discount100% off 3 days left at this price!
