
In recent days, millions of people have downloaded DeepSeek, just as they did with ChatGPT two years ago. But behind what may seem like just another chatbot, one that answers our questions, suggests recipes, organizses vacations, creates diet plans, and fitness programs, lies a much deeper revolution.
Artificial intelligence is much more than a chatbot: it is a set of general purpose technologies that are transforming our entire society. In particular, AI is reshaping:
- Medicine, by supporting early disease diagnosis and accelerating the discovery of new drugs.
- Public services, by making them more efficient and adaptable to citizens’ needs.
- The job market, by automating repetitive tasks and increasing productivity.
- Businesses, especially SMEs (small and medium enterprises), which can leverage AI to analyse data, enhance productivity, and compete more effectively in global markets.
But what exactly is happening with DeepSeek? Has it really outperformed American startups with a better business model? Did it copy OpenAI? Why did the Italian Data Protection Authority block it? And what does it mean for Europe?
Let’s find out!
1. What is DeepSeek, and why is it making waves?
DeepSeek is a generative artificial intelligence model developed by a Chinese startup founded in 2023 by Liang Wenfeng, a Chinese AI expert in his forties with a background in finance. DeepSeek works similarly to ChatGPT, generating text and answering questions. What surprised everyone is that it used far fewer resources than American giants like OpenAI and Google, yet its model operates at nearly the same level, and in some cases, has even outperformed ChatGPT.
This means two things:
- It is possible to compete with major American tech companies while spending significantly less.
- Despite U.S. chip restrictions, China has found alternative solutions, innovating through software and proving that it can compete with OpenAI and Google even without access to Nvidia’s most powerful chips.
🌍 Market Impact
DeepSeek’s success has spooked investors, causing an 18% drop in Nvidia’s stock value, the leading AI chip manufacturer, and wiping out $560 billion in market capitalization.

2. Has it really disrupted the American startup business model?
DeepSeek did not invent a new technology from scratch but combined existing methods in an innovative way, significantly reducing computational costs.
The key factor was the necessity to bypass U.S. restrictions on advanced chip exports, which forced Chinese researchers to find creative ways to work with less powerful chips, necessity is the mother of invention.
To better understand DeepSeek’s innovations, let’s look at some practical examples:
Mixture of Experts (MoE) – Hospital Analogy 🏥
Imagine going to a hospital for a health issue. If a single general practitioner had to handle every type of illness, it would take much longer to make accurate diagnoses.
Instead, hospitals have specialized doctors: a cardiologist for heart issues, a neurologist for brain disorders, an orthopedist for bone-related problems. Each patient is directed to the most suitable specialist, improving efficiency and accuracy.
➡️ DeepSeek applies the same principle: it activates only the most relevant “expert units” to answer a question, optimising the process and reducing resource consumption.
Load Balancing – Restaurant Analogy 🍽️
Imagine you own a busy restaurant. If only one chef prepared all the orders, customers would have to wait a long time, leading to dissatisfaction.
Instead, a well-organized restaurant distributes tasks among multiple chefs, each specializing in different dishes: one prepares pasta, another handles main courses, and another focuses on desserts. This balances the workload efficiently, reducing wait times and improving service.
➡️ DeepSeek uses the same principle: it distributes computational workload across different units, preventing bottlenecks and maintaining high performance even with less powerful hardware.
3. Did DeepSeek copy OpenAI?
OpenAI has accused DeepSeek of using distillation, a technique where a smaller model “learns” from a larger one by analysing its outputs.
Example: If a university student takes notes from a professor, summarizes the lectures, and creates a study guide for a classmate, that classmate learns the material without having to attend the full course.
Distillation is a common practice, but OpenAI’s terms of service prohibit users from “copying” its services or “using outputs to develop competing models.” It appears that DeepSeek used this technique to “learn” from OpenAI’s models through APIs. Some reports even claim that up until a few weeks ago, if users asked DeepSeek for its name, it would respond “ChatGPT”, suggesting it might have been trained using OpenAI generated data.
There is no definitive proof that DeepSeek used distillation illegally, but OpenAI and the U.S. government are investigating the matter.
4. Why did the Italian Data Protection Authority block it?
Italy was the first European country to investigate DeepSeek, just as it had done with ChatGPT.
The Italian Data Protection Authority (Garante per la Privacy) raised several critical questions: what personal data does DeepSeek collect? From which sources? For what purposes? On what legal basis? Are data stored in China? How are users informed about data usage?
n response to the inquiry, DeepSeek voluntarily withdrew its app from the Italian market and provided a generic statement to the regulator. However, the authority found the response insufficient and lacking transparency, leading to an immediate suspension of DeepSeek’s services in Italy.

Communication from the Italian Data Protection Authority: https://www.garanteprivacy.it/home/docweb/-/docweb-display/docweb/10097450
🚨 Key concerns:
- The authority wants to ensure transparency regarding user data collection and processing.
- The location of DeepSeek’s servers is a critical issue: if data is stored in China, it may be subject to government surveillance laws.
- DeepSeek may have used web scraping to gather data without consent, an issue already seen with other AI models.
The White House is also monitoring DeepSeek due to national security concerns. The U.S. fears that Chinese AI could collect sensitive data or be used for public opinion manipulation, particularly given China’s strict regulations and state control over tech companies.
5. Does it represent an opportunity for Europe?
Europe is still figuring out how to compete in the global AI race, while investing far less than the U.S. and China.

📉 Recent AI investments worldwide:
- 🇺🇸 USA: $500 billion in the STARGATE program
- 🇨🇳 China: $140 billion in its AI ecosystem
- 🇪🇺 Europe: only €2 billion through the AI Factory initiative
On 29 January, the European Commission presented a new plan “The Competitiveness compass” with promising AI initiatives, but Europe still invests far less than its global competitors.
It is not enough to just regulate AI, Europe must create a single digital market, reduce bureaucratic barriers, and increase funding for AI startups.
6. Europe at a Crossroads
This week, in response to DeepSeek and the U.S. STARGATE program, leading European AI organizations (ADRA, EAIF e EIT Digital) issued a joint appeal: Europe cannot afford to lag behind or rely on foreign AI providers.
They called for bold action to build a sovereign and competitive European AI ecosystem, as outlined in the Draghi and Letta reports. The message is clear: Europe must drastically increase investment in AI, simplify regulations, and foster public-private partnerships to accelerate homegrown AI innovation.

They called for bold action to build a sovereign and competitive European AI ecosystem, as outlined in the Draghi and Letta reports. The message is clear: Europe must drastically increase investment in AI, simplify regulations, and foster public-private partnerships to accelerate homegrown AI innovation.
📜 Link to the joint letter: European AI Call to Action
Conclusion: DeepSeek is just the beginning of a new AI era
DeepSeek is not just another chatbot, it is a wake-up call that AI is evolving faster than expected. Its sudden rise shows that AI is no longer monopolised by a few U.S. tech giants, but is becoming a global, decentralised technology capable of reshaping economic and geopolitical dynamics.
This AI revolution presents both opportunities and risks. To harness its potential, we must learn how AI works, understand its risks, and implement responsible safeguards.
One thing is certain: AI is advancing at an unprecedented pace, and those who fail to invest in or understand these technologies risk being left behind.
Back to all news