When Chinese AI startup DeepSeek released its open-source Large Language Model (LLM), DeepSeek V3, in December last year, it challenged the notion that cutting-edge AI development is exclusive to Big Tech firms with billions in funding and vast GPU resources.

In a technical paper, DeepSeek revealed that it built its model for just $5.6 million— a fraction of what OpenAI spent developing the models behind ChatGPT. Moreover, due to U.S. sanctions, DeepSeek lacked access to high-end NVIDIA GPUs. Despite this constraint, the company made another breakthrough last month with the release of DeepSeek R1, a reasoning model that reportedly surpasses OpenAI’s GPT-4o1 on multiple benchmarks.

The result? DeepSeek has captured global attention, sending shockwaves through Silicon Valley.

Here are the top 12 developments since:

1. Wall Street meltdown

This development from China generated significant excitement within the developer and open-source communities. However, among investors, it triggered major sell-offs, fueled by growing concerns over the ability of U.S. companies to compete with their Chinese counterparts in the AI arms race.

This led to dramatic drops in stock prices for major tech companies such as NVIDIA with a reported loss of over $1 trillion in market capitalization. NVIDIA, one of the companies gaining significantly since the GenAI boom, lost nearly $589 billion in market capitalization on a single day trading.

Other companies impacted include Oracle, Broadcom, Microsoft, etc. Interestingly, this was just a few days after the U.S. President Donald Trump announced Project Stargate, a $500 billion investment to boost the countries’ AI capabilities.

2. DeepSeek overtakes ChatGPT in the U.S. Apple App Store

Just a week after the release of its SOTA reasoning model, on January 26, the mobile application of DeepSeek replaced ChatGPT as the top free app on the Apple App Store in the U.S and 51 other countries. The data was shared by mobile app analytics firm Appfigures.

More than 80% of these downloads occurred in the last week. On Google Play Store, DeepSeek entered the list of the top 15 during the same time, rapidly climbing up from No. 133 on January 24.

3. OpenAI claims DeepSeek stole its data

According to Bloomberg and the Financial Times, Microsoft and OpenAI investigated whether DeepSeek improperly trained its R1 model using outputs from OpenAI models. OpenAI alleges that the Chinese startup might have used a process called distillation to incorporate OpenAI’s abilities to train its model.

According to OpenAI’s licensing terms and conditions, the outputs generated by its models like GPT-4 and others, can’t be used to train another model. This, however, sparked sharp criticism of OpenAI, given the company also scraped vast amounts of internet data to train its models.

Additionally, OpenAI and other companies developing similar AI models have faced lawsuits from artists, publishers, and writers over allegations of data theft.

4. Trump calls DeepSeek moment a “wake-up call”

Newly elected U.S. President Donald Trump stated that the rapid rise of DeepSeek should be a “wake-up call” for America’s tech companies. Even though he also expressed confidence that U.S. tech companies would continue to lead in AI, he acknowledged the growing competition from DeepSeek.

“The release of DeepSeek AI from a Chinese company should be a wake-up call for our industries to stay laser-focused on competing,” Trump reportedly said. He also said what DeepSeek achieved was a welcoming development for the world of AI.

Notably, just on his first day back in office, Trump scrapped the AI executive order signed by former President Joe Biden in October 2023.

5. Multiple countries block DeepSeek

On January 30, 2025, Italy became the first country in the world to block DeepSeek. The Italian data protection authority ordered the Chinese startup to block its chatbot in the country after it failed to address concerns regarding its privacy policy.

Earlier, the authority sought information from DeepSeek on how it handles personal data. It specifically requested details on the types of data collected, the sources of this data, the purposes for which it is used, the legal basis for its processing, and whether the data is stored in China.

Soon after, Australia also prohibited the use of DeepSeek on all government devices due to security concerns. Other European nations such as Belgium, France, and Ireland as well as South Korea also plan to seek similar data from DeepSeek.

Meanwhile, Taiwan, which has ongoing tensions with China, has banned the use of DeepSeek’s AI model in government operations, citing security risks and concerns about potential data exposure to China.

6. NASA and U.S. Navy ban DeepSeek

The National Aeronautics and Space Administration (NASA) became the latest U.S. government agency to block DeepSeek over security concerns.
In a memo shared with all personnel, NASA’s Chief Artificial Intelligence Officer (CAIO) David Salvagnini warned that DeepSeek’s servers are located outside the country, raising concerns over national security and data privacy.
Last month, the U.S Navy also warned its members to refrain from using DeepSeek’s AI model citing potential security and ethical concerns. During the same period, the state of Texas in the U.S. became the first state in the country to ban DeepSeek from any government use.

7. DeepSeek now available on Microsoft Azure and Github

Microsoft made DeepSeek’s models available on the Azure AI Foundry and GitHub model catalog, providing businesses with access to advanced AI solutions while ensuring compliance with security and reliability standards.

Interestingly, just days ago, Microsoft was investigating whether DeepSeek has used distillation to incorporate OpenAI’s abilities to train its model. Moreover, Microsoft has invested billions of dollars in OpenAI in the last few years.

8. Anthropic CEO questions DeepSeek claims, bats for export control

In a personal blog post, Anthropic CEO Dario Amodei stated that the claims about DeepSeek’s achievement at the cost of $6 million — an accomplishment that has cost American AI companies billions — are exaggerated. He claimed his startup has also developed its flagship model Claude 3.5 Sonnet at a significantly lower price nearly a year ago.

Additionally, he called for stricter export controls, specifically aimed at further restricting NVIDIA from selling its chips to China. Currently, the U.S. government prohibits NVIDIA from exporting its high-end chips to China. As a result, NVIDIA exports a customized version of its GPUs that deliver lower performance than the company’s most advanced models.

Amodei believes doing so would give the U.S. a significant advantage as both the countries lock horns to emerge as the leader in AI. “Well-enforced export controls are the only thing that can prevent China from getting millions of chips, and are therefore the most important determinant of whether we end up in a unipolar or bipolar world,” Amodei wrote.

9. Alibaba claims its model beats DeepSeek’s

DeepSeek is not the only company in China that is working on AI models. Chinese e-commerce giant Alibaba last month released a new version of its flagship AI model Qwen. The company claims Qwen 2.5-Max surpasses the abilities of DeepSeek V3.

“Qwen 2.5-Max outperforms… almost across the board GPT-4o, DeepSeek-V3 and Llama-3.1-405B,” Alibaba posted on its official WeChat account.

10. Trump meets Nvidia CEO, discusses DeepSeek

Last week, U.S. President met Nvidia CEO Jensen Huang at the White House to discuss exactly what Amodei proposed, over concerns of China’s growing dominance in AI, fuelled by DeepSeek.

Their discussion centered on reinforcing U.S. leadership in AI technology and tightening export restrictions on advanced AI chips to China. While President Trump refrained from sharing specific details, he characterized the meeting as “good.”

11. Altman doesn’t plan on suing DeepSeek

OpenAI CEO Sam Altman recently stated that he has no plans to sue Chinese startup DeepSeek. Last month, Altman’s company investigated if DeepSeek used ChatGPT to train its model.

“No, we have no plans to sue DeepSeek right now. We’re focused on building great products and leading the world in model capabilities, and I believe that will serve us well,” Altman told reporters. “We’re happy to have another competitor. We’ve faced many before, and I think it’s in everyone’s interest for us to keep advancing and maintaining our leadership.”

12. U.S. senator proposes $100 mn fine for downloading DeepSeek

In a new bill, Republican senator Josh Hawley from Missouri proposed to impose penalties of up to 20 years in prison, a $1 million fine, or both for individuals who knowingly download Chinese-developed AI models like DeepSeek.

If passed, the bill would prohibit anyone in the U.S. from using China-developed AI technology and similarly restrict the use of U.S.-developed AI in China.