DeepSeek And Its Impact On The Generative AI Global Race

Since January 20, 2025, launched the public, the Chinese startup Deepseek’s Chatbot with an open code The storm took the world of technology.

As a top -notch -free app on the American Apple App Store from January 26 – January – 16 million downloads of applications in their first 18 days (Chatgpt has had 9 million in the same time frame) – Deepseek’s performance and an accompanying search feature is at least the same compared to Openi’s Chatt for a fraction of the cost.

Its launch brought a US AI technology company, Nvidia The largest decline in market values ​​for an American company In American history history. That’s a pretty entrance!

American technical analysts and investors seem to be now lagging behind in the generative AI Global Race.

This can be justified considering how fast and economical Deepseek managed to develop R1 and get out of the door.

Deepseek uses Learning of reinforcementWhich means that the model learns complex behaviors of explanation by strengthening without controlled fine adjustment, which allows it to save significant computer resources.

But will Deepseek really appear as a leader in AI? And what are the consequences for this development for the future of searching? Let’s dive.

What happened since Deepseek started?

Although US technological companies were humiliated by speed and requested the cost of this launch, Deepseek’s arrival was not without controversy.

Many questions are lurking, ranging from suspicious violations of intellectual ownership to safety, privacy of data, Chinese censorship and the right costs of his technology.

Legal issues for copyright and data protection

Openi and Microsoft explore whether Deepsek used API Openi to integrate his AI models in Deepseek’s own models.

Openai claims to have evidence of Deepseek distilled out open to build a rival modelwhich is against Openai’s conditions of service, but probably is not against the law.

Distillation allows the transfer of a large pre -trained model to a smaller model, which allows a smaller model to achieve comparable performance with a great deal of cost.

This is more than a bit ironic given the Openai lawsuits due to ignoring the conditions of providing the service of another website and the use of their websites protected by the copyright for training of their systems.

There are also questions about where user data is stored and how are they processedGiven that Deepseek is a startup based in Chinese.

For anyone who manages customers’ data and payment details, an integration of tools such as Deepseek who stores the data aside the jurisdiction could violate the data protection laws and expose sensitive data to an unauthorized approach.

Considering that Deepseek has yet to provide their privacy policies, experts in industry and safety researchers advise the use of extreme caution with sensitive information in Deepsek.

Deepseek security violation

Wiz Research, a company specialized in cloud security, has announced that she is able to Hack deepseek and expose security risks with relatively ease of January 29.

She discovered a publicly available database that belonged to Deepsek, which allowed him to complete control of the database and access to the API user data and keys.

Wiz alerted the team of Deepseek and immediately took action to ensure the data. However, it is unclear who else has accessed or downloaded the information before being insured.

Although it is not uncommon for the startups to move and make a mistake, this is a particularly big mistake and shows Deepseek’s lack of focus on cyber -surely so far.

Concerned with national security similar to a tictic

There are worries about national security due to the rules of deepseek data collection Reminiscent of the fears for the Tictokwhich is a similar increase in global prominence from the Chinese Bytetlance.

The US government briefly banned Tiktok in January 2025, which came out of concern about the way the company collected users data. There were also fear that the Chinese government could use a platform to influence the public in the US

AND Few incidents in the last few years This has triggered that fear involves Tictok employees who use the information about the location from the journalist’s monitoring application to find a source of leaked information, and Tikict employees are allegedly plans for monitoring certain US citizens.

While Tictok is currently active in the US, his future has not been confirmed.

For similar reasons as a tictok concern, numerous governments around the world, including Australia and Italythey are already working on a ban on Deepseek from government systems and devices. AND The US also considers a ban on on Deepsek.

Chinese censorship

Whether you run deeply locally or in its app, Deepseeek Cenzura is present for inquiries that the Chinese government considers sensitive, According to the wiring investigation.

However, since it is an open source, there are ways to visit censorship, but it’s hard.

This would require launch on your own servers using a modified version of the publicly available Deepseek code, which means you will need access to several very advanced GPUs to launch the most powerful R1 version.

Questions about costs

Much has been written about the cost of constructing Deepseek. The initial statements of Deepseek were that It took under $ 6 million to build a construction Based on the price of NVIDIA’s rental GPU.

However, a The report of the half -conscious onesSemid research and counseling company has since claimed that Deepseek’s hardware consumption has been more than $ 500 million, along with additional research and development costs.

For context, Openai lost about $ 5 billion in 2024. And predicts that he will lose more than $ 11 billion in 2025. Even if Deepseek cost $ 500 million or more, it still reduced costs compared to what leading competitors spend.

So how did they reduce costs?

Before Deepseek came, the leading AI technology were built on neural networks, and these are mathematical systems that learn skills by analyzing huge amounts of data. This requires large amounts of computer power.

Specialized computer chips called graphics unit for processing (GPU) are an effective way for this type of data analysis. This is how Chip Maker Nvidia grew to prominent (and also had a huge decline in market values ​​on the day Deepseek was launched).

The GPUs cost about $ 40,000 and require a considerable electricity, which is why leading AI technologies like Openi Chatgpti were so expensive to build.

Sending data between chips may also require more energy than launching chips itself.

Deepseek managed to reduce costs, most of all Using a method called “a mix of experts. “

Instead of creating a neural network that learned data forms on the Internet, they divided the system into many neural networks and launched less “professional” systems paired with a “general” system, reducing the amount of data needed to travel between GPUs.

Implications of being an open code

Deepseek-R1 is both an “open pether” as any llm so far, which means that everyone can download, use or change their code.

Similar to the meta lami, codex and technical explanations They are sharedEnabling developers and organizations to use the model for their business needs, but the training information is not fully published.

Many believe that Deepseek is a big step towards democratization AI, allowing smaller companies and developers to build on Deepseek-R1 and achieve more AI features faster.

This could lead to multiple innovations in places with a limited approach to the technology needed to build AI solutions.

But critics are afraid that open code models can expose security vulnerability that could be used, which we have already seen in Deepseek’s first weeks in public.

Deepseek and the future SEO -A

So what does it really mean for search experts? The way I see it, Deepseek is just the next dirty Ai chatbot with search options in the fast variable SEO world.

It is important to understand that while tools such as deepseek and chatgpta use advanced natural language processing (NLP) and machine learning, they still simply give answers to the real questions that real people ask.

Their answers focus strongly on semantic understandingMatch intent and contextual analysis, but they ultimately serve the same need for the core of the user.

Although we have many years of experience testing the optimization tactics on established search engines like Google, we are still in the initial stages of understanding optimization for Generative ai chatbots.

Final thoughts

Whether Deepseek will be kept and grow in prominent, it remains to be seen.

Obviously, if other governments follow Australia, Italy and potentially now to ban Deepseek, it would limit his potential for growth.

And as much as Deepseek quickly became prominent, providing a draft for others and significantly reducing costs, the new AI that moves on the market could always be around the corner.

No matter what happens to Deepsek, we are at the beginning of a very fast period Innovation in AI technology.

As a SEO professionals, we need to be ready to test the rush of new platforms and the reverse engineer bringing their answers to user inquiries.

More resources:

---

Separate Picture: Phonlamai Photo/Shutterstock