Deepseek: China's AI revolution under the shadow of surveillance-severe allegations from Washington

Geopolitical tensions by AI: Why Deepseek's R1 model splits the world

The headlines of the past few months have been determined by a Chinese company that fundamentally changed the global AI landscape. Deepseek, a comparatively young startup from Hangzhou, not only attracted the attention of the technology world with its groundbreaking AI model R1, but also sparked violent controversy about data protection, espionage and geopolitical tensions.

A startup changes the rules of the game

Deepseek was only founded by Liang Wenfeng in 2023, a 40-year-old hedge fund manager without a traditional technology background. Wenfeng, who originally dates from the area of ​​quantitative finances, had already launched the HedgeGefonds High-Flyer together with two fellow students at the University of Zhejiang in 2015. This fund, which specialized in mathematical models and artificial intelligence for investment strategies, quickly developed into one of the leading quantitative hedge funds of China with an administration of over $ 13 billion.

What distinguishes Deepseek from other AI companies is its unusual financing structure. The company is financed exclusively by high-flyers and is therefore independent of external investors. This independence made it possible Wenfeng to acquire thousands of Nvidia processors in 2021-long before the US export restrictions for AI chips to China and to start an ambitious AI project.

Suitable for:

• China's catch -up in artificial intelligence: the case Deepseek and the strategic data use

Technical breakthroughs at a fraction of the costs

January 2025 marked a turning point in the global AI industry when Deepseek presented its R1 model to the public. Within a few days, the Deepseek app conquered the top position in the app stores and even displaced Chatgpt from Openai to second place. The reaction of the financial markets was dramatic: Nvidia shares lost 17 percent of their value on a single day, which corresponds to a historical loss of value of $ 589 billion.

The Deepseek-R1 model impresses with its technical specifications and performance. With 671 billion parameters in total, however, it activates only 37 billion parameters per round, which is made possible by an innovative mixture-of-experts architecture. This efficiency is reflected in the costs: Deepseek claims to have spent only $ 5.6 million for training his model-a fraction of the estimated $ 100 million that Openai has invested in GPT-4.

The performance of the model is impressive. In the Aime 2024 benchmark, Deepseek-R1 achieves an accuracy of 79.8 percent, and even 97.3 percent for mathematical tasks (Math-500). Particularly noteworthy is the ability of the system to “simulate thinking”, an approach in which the AI ​​articulates its thinking processes and gradually solves complex problems.

Open source as a competitive advantage

A crucial factor for deepseek success is the decision to publish the R1 model under an open source license. This license enables everyone to use the model for free, modify and even use it for commercial purposes. While the model itself is under a special Deepseek license that contains certain usage restrictions, the source code remains completely open.

This openness is in stark contrast to the proprietary approaches of Google, Openai and other western AI companies. Users can download the full model or smaller variants directly and operate on their own servers or even local computers. This flexibility, combined with the drastically lower usage costs-Deepseek only requires 14 US cent per million tokens, while Openaai calculates $ 7.50-in principle, existing business models in question.

Integration of an independent and cross-data source-wide AI platform for all company matters-Image: Xpert.digital

Ki-Gamechanger: The most flexible AI platform-tailor-made solutions that reduce costs, improve their decisions and increase efficiency

Independent AI platform: Integrates all relevant company data sources

• This AI platform interacts with all specific data sources
  • From SAP, Microsoft, Jira, Confluence, Salesforce, Zoom, Dropbox and many other data management systems
• Fast AI integration: tailor-made AI solutions for companies in hours or days instead of months
• Flexible infrastructure: cloud-based or hosting in your own data center (Germany, Europe, free choice of location)

• Highest data security: Use in law firms is the safe evidence
• Use across a wide variety of company data sources
• Choice of your own or various AI models (DE, EU, USA, CN)

Challenges that our AI platform solves

• A lack of accuracy of conventional AI solutions
• Data protection and secure management of sensitive data
• High costs and complexity of individual AI development
• Lack of qualified AI
• Integration of AI into existing IT systems

More about it here:

• AI integration of an independent and cross-data source-wide AI platform for all company matters

Serious allegations from Washington

However, Deepseek's success story has a dark downside that has increasingly focused on in recent months. A high-ranking official from the US State Department has raised serious accusations against the Chinese company to Reuters. These allegations include several critical areas and raise fundamental questions about the security and data protection of Deepseek users.

The central accusation is that Deepseek "knowingly has supported China's military and intelligence operations and will probably continue to provide it". This support goes far beyond the mere open source access to the AI ​​models and includes the active transfer of user data and statistics to Beijing's surveillance apparatus.

The claim is particularly explosive that Deepseek is mentioned in over 150 procurement documents of the Chinese People's Liberation Army and other military organizations. This indicates that the company not only forwards passive data, but is actively providing technological services for the Chinese military.

Suitable for:

• Artificial intelligence | Marketing tactics of American companies with AI Angstmacherei

Millions of user data in sight

The data protection concerns are particularly serious when you look at the range of Deepseek. The company records over 47 million users worldwide every day. According to US parliamentarians, Deepseek transmits data from American users via a “backend infrastructure” to China, which is connected to China Mobile, a state telecommunications giant.

While Chinese companies are obliged to provide data on request from the government under Chinese law, the accusation against Deepseek continues. The company is intended to proactively share this information with the Chinese authorities, which is raising significant data protection concerns worldwide for millions of users.

Umohnecke of US sanctions by mailbox companies

Another serious accusation concerns the alleged circumvention of US export restrictions. Deepseek is said to have tried to gain access to Nvidia's powerful H100 chips via mailbox companies in Southeast Asia, which have been under US export restrictions since 2022. Washington fears that China could use this technology for military purposes or a lead in the AI ​​race.

"Deepseek has tried to use mailbox companies in Southeast Asia to avoid export controls, and Deepseek is trying to access data centers in Southeast Asia to access US chips from afar," said the US official. These allegations raise questions about the actual costs and the efficiency of deepseeks models, since access to high -performance chips could have significant effects on training costs.

Nvidia has partially rejected that Deepseek uses H100 chips and explains that the company only uses the weaker H800 chips that are approved for export to China. However, Reuters and other media were not able to verify this information independently.

The context of the global monitoring apparatus

The allegations against Deepseek must be viewed in the context of global intelligence activities. China has significantly expanded its surveillance capacities in recent years. The National Secretary Service Act adopted in 2017 gives the Chinese security authorities extensive powers to oblige individuals, companies and organizations at home and abroad. The anti-espionage law revised in 2023 significantly expanded the definition of espionage and created a tightened instrument for control and monitoring.

At the same time, it is important to note that western secret services also operate extensive surveillance programs. The NSA programs such as Prism and the British Tempora program revealed by Edward Snowden show that mass surveillance of internet and telephone communication is by no means a purely Chinese phenomenon. The NSA has access to large parts of internet traffic that goes across the American area and processes these amounts of data in its own systems. The British GCHQ can access 40 billion individual content data per day through the Tempora program.

Mutual manipulation and disinformation

The accusations against Deepseek should also be considered against the background of intensive geopolitical rivalry between the USA and China. Secret services of different nations have a long history of mutual manipulation and disinformation. The United States has repeatedly tried to limit China's progress in AI technology, while China on his part is trying to achieve technological independence.

The fact that relevant evidence of the allegations raised against Deepseek is not publicly exemplified should lead to a careful assessment of the situation. It could also be badly default by a direct state competitor, as the original text already indicated. The risk of cyber attacks and disinformation campaigns that aim to undermine trust in competing technologies is real and documented.

Technological sovereignty and innovation

Deepseek's rise is also to be understood in the context of China's striving for technological sovereignty. The US export restrictions on semiconductor chips should brake China's AI development, but may have caused the opposite and promoted innovation through restrictions. Brian Colello, a tech analyst at Morningstar, commented: "These Chinese models were boundary of processor, which led to some creative techniques during training, and the deepseek model has shown in view of the processors on which it was trained".

The efficiency of Deepseek shows that technological breakthroughs do not always depend on the greatest investments or the latest hardware. The company has proven that innovative approaches such as the mixture-of-experts architecture and advanced training methods can lead to impressive results, even if less powerful hardware is available.

The future of AI development

Deepseek's success has changed the AI ​​industry sustainably and raised important questions about the future of technology development. The combination of open source approach, drastically reduced costs and comparable service questions the business models of established companies. At the same time, the data protection and security concerns show the challenges associated with the increasing globalization of AI technologies.

The company is now considering external financing to cope with its rapid growth. Potential investors such as the Alibaba Group and Chinese state -related funds have already signaled interest, but this could increase Washington's concerns and restrict future business opportunities in the United States.

Data protection in the digital age

The controversy around Deepseek illustrates the complex challenges of data protection in the digital age. While China has had a comprehensive data protection law with the Personal Information Protection Law (PIPL) since 2021 that takes over the principles from the European GDPR, government agencies are largely excluded from these regulations. This creates a area of ​​tension between private data protection and state security interests.

The challenge is that users worldwide want to benefit from the technological advantages of innovative AI systems, but at the same time want to see their privacy and data security guaranteed. Deepseek's free offer and superior performance make it attractive for millions of users, but the concerns about data transfer to Chinese authorities are entitled and serious.

Suitable for:

• Artificial intelligence: BMW relies on China-Ki with Deepseek

Innovation between progress and risk

Deepseek represents both the enormous possibilities and the considerable risks of modern AI development. The company has shown that technological innovation is not limited to certain countries or companies and that creative approaches can lead to groundbreaking results. The open source philosophy and the drastic reduction in costs have the potential to democratize AI technologies and make it accessible to a wider user base.

At the same time, the Deepseek case illustrates the need for a differentiated view of technology companies from different countries. While the allegations of the US government have to be taken seriously, they should also be viewed in the context of global intelligence activities and geopolitical rivalries. The fact that Western secret services also operate extensive surveillance programs should lead to a balanced discussion about data protection and privacy that is not dominated by national interests.

The future of AI development will significantly depend on how well the international community succeeds in creating a framework that promotes innovation, but at the same time protects the rights and safety of the users. Deepseek has made an important contribution to this discussion by showing the possibilities and limits of current AI technology. The challenge now is to use these findings in order to establish responsible and ethical standards for the further development of artificial intelligence.

B2B procurement: supply chains, trading, marketplaces & AI-supported sourcing with accio.com-Image: xpert.digital

More about it here:

• Find products and B2B insights with AI / advice and support

☑️ SME support in strategy, consulting, planning and implementation

☑️ Creation or realignment of the AI ​​strategy

☑️ Pioneer Business Development

Konrad Wolfenstein

I would be happy to serve as your personal advisor.

You can contact me by filling out the contact form below or simply call me on +49 89 89 674 804 (Munich) .

I'm looking forward to our joint project.

Xpert.Digital is a hub for industry with a focus on digitalization, mechanical engineering, logistics/intralogistics and photovoltaics.

With our 360° business development solution, we support well-known companies from new business to after sales.

Market intelligence, smarketing, marketing automation, content development, PR, mail campaigns, personalized social media and lead nurturing are part of our digital tools.

You can find out more at: www.xpert.digital - www.xpert.solar - www.xpert.plus