A mysterious Chinese startup has taken the AI world by surprise, shaking up big tech companies in Silicon Valley, causing waves in global stock markets and redefining what people thought AI was capable of doing. DeepSeek combines huge funding, bringing together a large amount of financial resources, or investment to support its operations, growth and goals; a commitment to open-source innovation; and a strong goal of exceeding human intelligence—all the while outperforming such big names as OpenAI. However, its fast rise also raises some important questions.
However, such fast growth raises questions: Are we truly seeing an AI revolution, or is it just overhyped? In the next sections, we will take a closer look at DeepSeek’s origins and mission; see how its models compare to such well-known AI tools as ChatGPT; explore the major market shifts it has caused; and examine privacy concerns similar to those linked to TikTok.
What is DeepSeek & who is in charge?
DeepSeek was started in July 2023 by Liang Wenfeng, a well-known graduate of Zhejiang University. Based in Hangzhou, the company is strongly supported by both funding and guidance from High-Flyer, a quantitative hedge fund—a type of investment company that uses maths, data and computer programs to decide where to invest money and make profits—also co-founded by Liang.
With strong financial support from High-Flyer, a company managing around $8 billion in assets, DeepSeek has made a powerful move into the AI industry. It has managed to secure a large number of Nvidia A100 chips, even though their export to China is restricted. These chips are essential for powering the company’s technology and driving innovation.
DeepSeek became well-known for creating its DeepSeek-V3 model, built with just $6 million in computing resources—or the tools and technology, such as powerful computers, servers and chips needed to process large amounts of data and run complex programs. This is minuscule compared to what big US tech companies usually spend on similar projects.
Thanks to its efficiency, DeepSeek’s AI Assistant has become the most popular free app on the US App Store, even beating ChatGPT. This success shows that DeepSeek can deliver great results at lower costs, disrupting the market—or shaking up the usual way things are done—and pushing the global AI industry to rethink its approach.
DeepSeek’s technology is powered by a research team from High-Flyer, which announced in early-2023 that it would focus on developing Artificial General Intelligence (AGI)—AI that can think, learn and solve problems like a human. Unlike regular AI, which is designed for specific tasks, AGI can handle a wide range of tasks and adapt to new situations on its own.
This research team, which operates a network of 10,000 A100 chips, is working to push AI beyond its usual uses, aiming to create systems that can outperform humans in important, and value-added, tasks. By combining its resources effectively, DeepSeek shows a strong commitment to being a leader in AI. This focussed strategy could have a big impact on the future of AI.
Story about the Origin of DeepSeek
DeepSeek was born out of the excitement, and competition, that started when OpenAI launched ChatGPT in late-2022. This sparked a race among Chinese tech companies to create their own advanced AI chatbots, pushing innovation forward. Although such major companies as Baidu, a leading Chinese technology and search engine company, made early attempts, it became clear that there was a noticeable gap between US and Chinese AI technologies.
This left many in China’s tech community feeling disappointed. This situation led to the creation of DeepSeek in 2023, with a clear goal of closing this gap. Its mission was to push the limits of AGI and create AI that could go beyond human intelligence.
DeepSeek was founded with a focus on idealistic goals rather than making money, much like OpenAI in its early years. They saw themselves as a team of dreamers, motivated by the desire to make big breakthroughs in AGI, not by the goal of making money. This vision is supported by significant investments in technology, especially in creating their advanced models, DeepSeek-V3 and DeepSeek-R1.
These models have quickly earned praise for their impressive performance, competing with—and even surpassing—some of the top models from OpenAI and Meta. This is even more remarkable because the company has limited access to the newest Nvidia chips. DeepSeek-V3 stands out for its fast processing speed and cost efficiency. It has made great progress in areas that need heavy computing power, such as coding and solving complex math problems.
A research paper on Arxiv highlighted the model’s impressive abilities, showing that it outperformed other open-source models and performed on a par with such top closed-source models as GPT-4 and Claude-3.5-Sonnet. This achievement was also reported by the well-known science platform, InterestingEngineering.com.
How Do DeepSeek & ChatGPT Differ?
DeepSeek sets itself apart from such AI tools as ChatGPT by using a different design and way of working. These differences aim to make it work faster and cost less to run.
Using a system, called ‘Mixture of Experts’ (MoE), DeepSeek activates only the necessary parts of its network for each question. This approach helps save computing power and lower costs. Unlike DeepSeek, ChatGPT uses a transformer-based design that runs tasks through its entire network, which requires more resources and increases costs.
Tech & Performance Advancements
As said earlier, DeepSeek’s MoE design focusses on processing tasks based on their specific needs. This improves its performance in such areas as coding and solving technical problems, while also making responses faster. On the other hand, ChatGPT provides steady performance across many tasks, but it can be slower because it processes everything through its entire system.
Even so, ChatGPT, usually, gives more detailed and thoughtful answers, offering a depth that DeepSeek may miss when dealing with broader topics.
About Personalization/Applications
DeepSeek’s system is built to be highly flexible, making it ideal for users with specific needs who have the technical skills to adjust its features. This makes it a great fit for specific tasks that need accurate technical support.
In contrast, ChatGPT stands out with its easy-to-use customization features, making it a great choice for a wide range of people. It is especially useful for such tasks as creative writing, coming up with ideas and finding general information.
Talking about Price and Availability
When it comes to accessibility, DeepSeek is open-source. It means that it is completely free to use and can be modified easily, which makes it popular among developers. On the other hand, ChatGPT offers a free version, but also has paid plans that unlock advanced features and better API access, which means allowing different software, or apps to connect—and work with—a tool or service. It is like giving other programs a way to ‘talk to’, and use, the features of the tool.
User Experience and Ethical Factors
DeepSeek is a powerful tool, but it requires more technical skills, which may make it harder for people without a tech background to use. In contrast, ChatGPT has a simple, and user-friendly, design, making it easier for anyone to learn and use.
DeepSeek raises ethical concerns, because it collects such data as IP addresses and device details, which may not fully comply with such privacy rules as the General Data Protection Regulation (GDPR), considered the world’s strongest privacy and security law—a European Union (EU) law that protects individuals’ personal data—which was adopted in 2016 and came into effect in 2018. On the other hand, OpenAI focusses on protecting user privacy by using data encryption and keeping information anonymous to better follow privacy laws.
What Does DeepSeek’s Success Imply?
The release of DeepSeek’s R1 model has caused major waves in global stock markets, especially in the tech industry. On one of the important trading days, the Nasdaq Composite index dropped by 3.1%, wiping out more than $1 trillion in market value. CNBC reported that this market drop was driven largely by losses in big tech companies. Nvidia was hit the hardest, losing over $700 billion in value and suffering the biggest single-day loss ever recorded for a company.
Disruptive Innovation & Nvidia Tech
Some plausible reasons why Nvidia may experience such a loss in this fictional situation:
– Superior Performance: DeepSeek’s R1 model could offer significantly better performance in AI tasks compared to Nvidia’s GPUs, making them less desirable for AI developers and researchers
– Cost-Effectiveness: R1 could be more cost-effective than Nvidia’s offerings—either in terms of initial purchase price, or operating costs—making it a more attractive option for budget-conscious users
– Specialized Hardware: DeepSeek may have developed specialized hardware specifically optimized for running R1, which could outperform general-purpose GPUs from Nvidia in AI applications
– Disruptive Innovation: R1 could introduce a disruptive innovation in AI technology, such as a new architecture, or algorithm, that renders Nvidia’s existing GPU technology less competitive
– Market Share Loss: If R1 gains widespread adoption, Nvidia could lose significant market share in the AI hardware market, leading to a decline in revenue and investor confidence
– Shift in Demand: R1’s capabilities could trigger a shift in demand towards specific AI applications, or workloads better suited to DeepSeek’s hardware, reducing the demand for Nvidia’s GPUs
– Negative Sentiment: The release of R1, and its potential impact on Nvidia, could create negative sentiment among investors, leading to a sell-off of Nvidia’s stock and a decline in its market capitalization
It is important to remember that this is a fictional scenario. The actual impact of any new AI model on Nvidia would depend upon various factors, including the model’s capabilities, pricing, market adoption and Nvidia’s own response and innovation.
DeepSeek Beyond the Stock Market
This change made Apple the most valuable company in the US, pushing Nvidia out of the top spot. Other tech giants, such as Google and Microsoft, saw significant losses, too.
– DeepSeek’s steady progress is affecting more than just stock prices. The energy sector took a hit as investors grew worried that DeepSeek’s energy-efficient technology could reduce the tech industry’s overall energy needs
– Stocks for such companies as GE Vernova and Vistra, which make wind and gas turbines, dropped sharply by 21% and 28%, respectively
– Wall Street experts are carefully examining the long-term impact of DeepSeek’s rise as a strong player in the AI industry. DeepSeek’s cheaper, and more energy-efficient, models are raising concerns on whether US companies are spending too much on AI technology, pointing to a possible overspending issue in the sector
– Geopolitical issues are also a factor, with uncertainty around how the US government could respond. Questions about tariffs on Chinese imports and new regulations could impact how such technologies as DeepSeek are adopted and used in US markets
Pertinent: Is Your Privacy at Risk?
DeepSeek, even with its advanced technology, is facing criticism over privacy concerns, similar to the issues raised about other Chinese-owned platforms, such as TikTok. People are mainly worried because DeepSeek gathers a lot of personal information, such as birth dates, typing patterns, text and voice inputs, uploaded files and chat history, all of which are stored on servers in China.
This raises serious concerns about user data security and privacy—especially since China’s strict intelligence laws require all organizations to assist with national intelligence activities.
Adding to the concerns, such popular AI models as ChatGPT and Google Gemini have raised a red flag about DeepSeek. They have pointed out potential risks, especially because the technology comes from China, which is a sensitive issue given the current global political tensions.
People are worried that the software may have hidden viruses, or spying tools built into it, which could threaten user safety. DeepSeek’s security came under scrutiny after a flaw was discovered in December. This weakness could have allowed hackers to take over accounts, using prompt injection. However, the issue was later fixed.
The issue worsened when DeepSeek revealed that its AI assistant had been hit by a “large-scale malicious attack” during a period of increased usage. This highlights ongoing security problems and raises concerns about protecting personal data handled by AI.
People using such AI tools as DeepSeek need to think about the privacy and security risks involved. The worries are not just about keeping data private, but also about how the collected information might be used without the user knowing. This could include training AI models, or other hidden activities. As AI continues to grow, it is important to have clear practices and strong security to protect personal information and not break users’ trust.
About Some Responses to DeepSeek
Nvidia acknowledged DeepSeek as a major step forward in AI, praising its use of test-time scaling—a method in AI where adjustments, or improvements, are made to the model while it is being used, without changing its original design, which helps the AI adapt better to specific tasks, or situations, during its operation. This approach enables the development of new models that follow export regulations.
Nvidia praised DeepSeek, but also noted that AI tasks depend a lot on Nvidia Graphics Processing Units (GPUs)—powerful computer chips made by Nvidia, originally designed for rendering images and videos in games, which are now widely used for such tasks as running AI programs, processing large amounts of data and handling complex calculations quickly and efficiently—and advanced networking. This highlights the need for powerful hardware to keep AI running smoothly.
Sam Altman, CEO of OpenAI, praised DeepSeek’s R1 model for delivering great results while being cost-effective. Altman highlighted that OpenAI was dedicated to advancing its research and expanding its computing power to reach its goals. While he acknowledged DeepSeek as an important achievement, he made it clear that OpenAI was staying focussed on its long-term plans.
The Chinese state media and the country’s politicians have shown a lot of interest in DeepSeek’s success. They see it as a way to challenge US leadership in technology and move closer to China’s goal of being self-reliant in AI. According to Reuters, DeepSeek’s founder took part in a major meeting with Premier Li Qiang, showing how important DeepSeek is to China’s national goals.
US President Donald Trump called DeepSeek’s growth both a challenge and an opportunity for the US tech industry. He said it should motivate American companies to innovate, and compete better, on a global level, pointing out the importance of its impact on both politics and the economy.
(The author of this article is an award-winning Science Writer and a Defence, Aerospace & Political Analyst based in Bengaluru. He is also Director of ADD Engineering Components, India, Pvt. Ltd, a subsidiary of ADD Engineering GmbH, Germany. You can reach him at: [email protected])
Link to article –
