What is DeepSeek? In Their Own Words

DeepSeek is a Chinese AI startup that made waves in January 2025 with the release of an advanced chatbot. This new AI, developed by DeepSeek, is said to rival OpenAI’s ChatGPT—but at a fraction of the cost. DeepSeek’s rapid rise is sending shockwaves through the AI industry.

Founded in 2023 by Liang Wenfeng, a Chinese hedge fund manager, DeepSeek quickly gained attention for its innovative approach. In January 2025, the company unveiled DeepThink R1, a “reasoning model” that displays its “thought process” to help users verify the AI’s responses. This transparency sets it apart from many other AI systems.

To test DeepSeek’s capabilities, I asked it to tell me a joke. Instead of relying on a well-known punchline, the AI crafted its own pun based on the phrase “puddle of clues.” The result? A playful joke about a Shih Tzu detective:

“Why did the Shih Tzu become a detective? Because he was a little poo-dle of clues!”

You can see the AI’s “chain of thought” in the screenshot below, showcasing its ability to generate original content.

One of the most remarkable aspects of DeepSeek’s success is its resourcefulness. The company built its large language model (LLM) using older, cheaper computer chips acquired before President Biden’s 2022 export ban on advanced semiconductors. By leveraging these cost-effective chips, DeepSeek managed to match—or even surpass—its competitors at a significantly lower cost. While ChatGPT reportedly cost $100 million to develop, DeepSeek achieved similar results for just $5 million.

The news of DeepSeek’s breakthrough has had a profound impact on the stock market. Nvidia, a leading supplier of AI chips, saw its stock plummet by nearly $600 billion in a single day as investors reassessed the future costs of AI development. Additionally, DeepSeek’s success has cast doubt on the feasibility of President Trump’s proposed $500 billion Stargate AI program.

However, DeepSeek’s rise has also raised concerns about privacy and security, particularly given its Chinese origins. While the platform itself may pose potential risks, its “open-model” architecture allows others to easily replicate and build similar tools, further complicating the landscape of AI development.