DeepSeek, a small Chinese tech firm established in 2023, succeeded in erasing nearly $600 billion from Nvidia, the world’s foremost tech company, representing the largest decline in US stock market history. DeepSeek — which reportedly employs less than 200 people — sent shockwaves through Wall Street and Silicon Valley with the launch of its new […]
DeepSeek, a small Chinese tech firm established in 2023, succeeded in erasing nearly $600 billion from Nvidia, the world’s foremost tech company, representing the largest decline in US stock market history.
DeepSeek — which reportedly employs less than 200 people — sent shockwaves through Wall Street and Silicon Valley with the launch of its new AI chatbots, DeepSeek V3 and R1.
After launching on Monday, the free chatbot quickly became the top download on Apple in both the US and UK, placing the startup in direct competition with ChatGPT and Google’s Gemini.
So, what is DeepSeek, and what do you need to know about it? Here are five key takeaways and reactions to the latest AI player.
What is DeepSeek?
DeepSeek was a startup founded in 2023 in China’s Hangzhou region by hedge fund entrepreneur Liang Wenfeng, who previously led the High-Flyer hedge fund.
Despite US efforts to restrict Chinese access to AI chips, the startup had accumulated a stockpile of Nvidia’s A100 graphics processors before Joe Biden implemented a ban and developed an AI model.
Last week, a research paper about the firm’s R1 model found it could perform at similar or even higher levels than some of the top AI models on the market, but it was also significantly cheaper.
Is this AI’s “Sputnik moment”?
In the 1950s, the US and Soviet Union were locked in a Cold War that extended beyond our planet and into space. In 1955, the US aimed to launch an artificial satellite by 1958, but the Soviet Union launched Sputnik 1 on October 4, 1957, becoming the first artificial Earth satellite. This sparked public fear in the West about the technological gap, leading to the Space Race that culminated in the 1969 Moon Landing.
Today, the US is widely regarded as the global leader in AI, home to prominent providers such as OpenAI, Google, Microsoft, X, and others. However, the launch of DeepSeek R1 has raised significant concerns.
The Chinese chatbot wiped $11 trillion from the US tech index, causing the likes of Nvidia, Google, and Microsoft to suffer.
In a post on X, a US venture capitalist compared the launch of DeepSeek’s R1 model to the launch of Sputnik.
Deepseek R1 is AI’s Sputnik moment.
— Marc Andreessen 🇺🇸 (@pmarca) January 26, 2025
Harder, Faster, Cheaper, Stronger?
When it launched R1 on Monday, DeepSeek made major claims about its model, saying it outperforms OpenAI’s o1-mini model “across various benchmarks”.
Independent research from Artificial Intelligence, which tracks AI models, found that DeepSeek AI outperformed Google, Meta and Anthropic models.
Compounding worries in Silicon Valley is the cost, with DeepSeek claiming it is also much cheaper to operate than OpenAI’s o1.
According to the firm, R1 costs just $5.6 million to develop, which compares favourably to Western models.
Anthropic co-founder Dario Amodei claims it can cost between $100 million and $1 billion to train new advanced AI models — far more than the investment in R1.
Goldman Sachs has already raised concerns about AI’s affordability for enterprises, questioning whether potential trillions invested in the technology will “ever pay off.”
DeepSeek has made its model available for free on its own infrastructure and released the source code, confirming its ability to match or perform better than flagship OpenAI or Meta AI models.
Adrian Sowa, senior consultant at Pentest People, told TechInformed that making it available for free “is part of a clever marketing strategy” but warned it may not be sustainable.
“Most likely, the free availability of such a powerful model will not be feasible long term in terms of costs. Thus, I would expect that DeepSeek will implement some kind of payment options in the future or adopt a similar approach to OpenAI’s ‘mini’ models.”
Instant target for cyber threats
Despite only launching its free assistant this week, DeepSeek has already been subject to a cyberattack, which caused the startup to temporarily limit registrations on its platform.
The firm found its new app had been targeted with a “large-scale malicious attack” on Monday. It has now reopened registrations.
Erich Kron, security awareness advocate at KnowBe4, says that availability is a key tenet of cybersecurity, so it should be no surprise that hackers immediately targeted DeepSeek after it made its app live.
“With the popularity of DeepSeek growing, it’s not a big surprise that they are being targeted by malicious web traffic. These sorts of attacks could be a way to extort an organisation by promising to stop attacks and restore availability for a fee, it could be rival organisations seeking to impact the competition negatively, or it could even be people who have invested in a competing organisation and want to protect their investment by taking out the competition.
“The cybersecurity world has become global, with attacks originating from any continent and targeting any organisation with a web presence. Unfortunately, many counter moves, such as pausing new user registration to allow computing resources to be freed up for other services, can bring back the use of the platform for some but also make for a bad experience for potentially new subscribers and can be very damaging to the organisation.
“In a time when internet outages can impact organisations to the tune of millions of dollars lost per hour, the threat of attacks such as this is very real and should be carefully considered and planned for.”
Deep mistrust: the great firewall of China
In recent years, trust in Chinese tech has hit a noticeable low. The US decision to ban Huawei from its telecoms infrastructure — followed closely by similar decisions in other Western countries — highlighted concerns about political influence on Chinese technology providers.
There is no evidence to suggest that the Chinese government has any influence over the AI startup — and it is important to note that Huawei has denied all accusations of impropriety — but its sudden burst onto the AI scene does raise some interesting questions about where we get our technology.
One key concern is censorship. In testing DeepSeek, the AI answers questions quickly and in detail, but several users have reported instances of the chatbot refusing to answer politically sensitive questions.
Operating out of China means the AI must not contain content that violates the country’s “core socialist values”.
According to the Guardian, DeepSeek refused to answer questions on subjects such as Tiananmen Square, the Umbrella Revolution, and the sovereignty of Taiwan.
While users have reportedly found workarounds, including switching letters for similar numbers, the issue still highlights why users should proceed with caution.
It is important to note that the Guardian study found similar results on some questions when put to Google’s Gemini AI, though ChatGPT seemed to answer most questions directly.
Darren Guccione, CEO and co-founder of Keeper Security, warned: “Leveraging AI platforms like DeepSeek may seem like a step forward, but organisations must carefully consider the risks — particularly when these platforms operate within regulatory environments where data access and oversight are less transparent.
“Inputting sensitive company information into these systems could expose critical data to state-controlled surveillance or misuse, creating a Trojan Horse into an organisation and all of its employees.
“This also has significant BYOD (Bring Your Own Device) security and user-level implications since employees might download this application on their personal device, which may then be used to transact on the organisation’s website, applications, and systems. This risk could have an exponential impact on data privacy and security if not properly managed.”
Reaction to China’s AI play
US President Donald Trump has been taking a significant interest in AI lately, last week announcing a new $500 billion AI venture called Stargate as part of a collaboration with OpenAI, Softbank, and Oracle.
Trump said the launch of DeepSeek should be a “wake-up call for our industries that we need to be laser-focused on competing to win.”
However, perhaps surprisingly, he also praised the model, saying that the idea that it can deliver the same performance as existing AI models at a lower cost is “a positive.”
“That’s good because you don’t have to spend as much money,” Trump said.
OpenAI boss Sam Altman also welcomed the launch of DeepSeek, saying he had been impressed by the R1 model.
“DeepSeek’s R1 is an impressive model, particularly around what they’re able to deliver for the price,” he said. “We will obviously deliver much better models, and also, it’s legit invigorating to have a new competitor. We will pull up some releases.”
But Tesla boss Elon Musk — who has been pushing his own AI, Grok, through his social media platform, X — appears to be sceptical of claims about DeepSeek.
Responding on X to a video of Scale AI CEO Alexandr Wang, in which Wang claimed DeepSeek actually used significantly more Nvidia chips than it had revealed, Musk tweeted, “Obviously”.
Obviously
— Elon Musk (@elonmusk) January 27, 2025
