There’s been a lot of debate over the last few days around DeepSeek R1– the Chinese competitor to ChatGPT that was developed for a fraction of the cost and quickly became the number one app download.
For those not keeping up with the news – financial markets reacted dramatically to the chatbot launch, with tech stock prices plummeting, including Nvidia, which fell by 17% in one day.
Fundamentally, this was because DeepSeek’s arrival immediately raised questions about the huge sums of money that everyone has always assumed is needed to develop AI.
If DeepSeek really was built for just $6 million, can the enormous trillion-dollar valuations of U.S. companies – not to mention the recent $500 billion Stargate investment announced by Donald Trump – really be justified?
Well, digging into the story, I quickly found that there’s a little more to it than was immediately apparent.
So, here’s my overview of the tech story that “crashed” the AI market, and what it could mean for the ongoing AI revolution.
What Exactly Is DeepSeek?
So basically, DeepSeek is an LLM-powered natural language chatbot (just like ChatGPT) developed by a Chinese company (also called DeepSeek).
It’s reportedly close to ChatGPT in terms of power – which is impressive considering that it is said to have been built for a cost of just $6 million. ChatGPT’s creator, OpenAI, has never been specific about how expensive its own chatbot was to build, but the company has attracted billions of dollars in investment.
Just like ChatGPT, DeepSeek can take a prompt and use it to create just about anything that can be written by humans, from poetry to computer code. Like all large language models (LLMs) it can do this because it’s been trained on huge amounts of text (this is the expensive part of building AI).
Where DeepSeek differs from ChatGPT, however, is that its creators have made it open-source. This means anyone can use, view or modify the code in any way they want. ChatGPT, by comparison, is proprietary software, and the exact nature of its inner workings remains a closely guarded secret.
Interestingly, it is this open element that gives us a clue about how it was created so cost-effectively.
AI pioneer and Meta chief scientist Yann LeCun pointed out that, following standard open-source practice, much of DeepSeek is actually built on top of existing, freely available AI code, such as Meta’s Llama LLM models.
It’s highly possible that this dramatically reduced the cost of training the DeepSeek LLM. Even so, news of its release still caused the biggest crash in tech stocks’ value in recent years.
Is AI Overpriced?
All of this has inevitably led to questions about whether the gargantuan sums of money being thrown at AI companies – particularly U.S. companies – can really be justified.
As we’ve already seen, these are questions that could have major implications for the global economy.
U.S. stock markets have been on a continuous upward trajectory for most of the past decade. Much of this growth has been driven by tech stocks, particularly by the belief that huge amounts of value will be generated by their investments in AI.
U.S.-based AI investors have also been caught off guard by the fact that DeepSeek’s accomplishments have come about despite not having access to the latest Nvidia AI processing technology. Nvidia’s most powerful AI processors are subject to a U.S. government ban on exports to China.
This means that DeepSeek has been forced to rely on performance-capped versions of the chips approved for export.
However, some believe that rather than being a barrier, this may actually have been a catalyst for innovation.
As reported by MIT Technology Review, “Rather than weakening China’s AI capabilities, the sanctions appear to be driving startups like DeepSeek to innovate in ways that prioritize efficiency, resource-pooling, and collaboration.”
The Privacy Question
Although it’s very quickly become very popular, it’s clear that DeepSeek does raise some red flags regarding privacy. In particular, these relate to the fact that it sends data back to its Chinese parent company.
DeepSeek makes no secret of this, meaning there’s no legal issue or potential breach of data laws like GDPR.
Just as with US-based chatbots, including ChatGPT and Google’s Gemini, users agree that they consent to their data being shared when they sign up for the service.
The difference, however, is that, unlike U.S. companies, Chinese companies are free, or sometimes even obliged, to share any information they collect with their government.
This could lead to the government of China – a leading contender in the global AI race – potentially getting access to huge amounts of Western citizens’ personal data.
Just as with the ongoing TikTok controversy, it boils down to fears that this data could give them an advantage when it comes to further training of AI systems.
To press home the fact that this is an issue of geopolitical significance, DeepSeek was apparently hit by a huge cyberattack very shortly after it was launched. Although no culprits have been identified as of writing, it’s claimed that it was a distributed denial of service (DDoS) attack, a form of attack primarily intended to take the service offline.
A New AI Era?
From a business point of view, though, DeepSeek’s triumphant launch is as much a victory of open source over closed, proprietary methods of AI development as it is of East over West.
By working within an ecosystem where the sharing of ideas and information comes before the need to generate profits, we’ve seen that innovation is likely to flourish.
As LeCun puts it, “Because their work is published and open source, everyone can profit from it. That is the power of open research and open source.”
Even Donald Trump, fresh from announcing the half-billion investment, struck an optimistic note when addressing the subject. If a Chinese company can create powerful AI tools at a fraction of the expected cost, then so can anyone else, he reasoned.
So, are we moving past the era when building AI tools is only possible for extremely well-funded global corporations and towards a more democratized development landscape?
Let’s hope so. Although I’m sure that giants like Microsoft and Google will continue to dominate the cutting-edge, open source and its community of collaborative, innovative builders will also play an important part. And if we’re able to put aside geopolitical issues and focus on moving forward together, that’s likely to be a good thing for everyone.
