DeepSeek, the Chinese AI Startup That Shook the Tech World
A remarkably efficient and powerful Chinese AI model has disrupted the technology industry, drawing global attention and shaking confidence on Wall Street. The AI model, named DeepSeek R1, was developed by DeepSeek, a startup founded just a year ago, and is being hailed as “AI’s Sputnik moment” by tech investor Marc Andreessen.
R1 has demonstrated capabilities comparable to those of established AI giants such as OpenAI’s GPT-4, Meta’s Llama, and Google’s Gemini — but at a fraction of the cost. DeepSeek revealed it spent just $5.6 million to power its base model, a shocking contrast to the billions invested by leading U.S. tech companies. This feat was achieved despite U.S. export restrictions on high-power AI chips, which have aimed to curb China’s technological advancements.
DeepSeek was founded in late 2023 by Liang Wenfeng, a former hedge fund manager who has become an AI evangelist in China. His hedge fund, High-Flyer, focuses on advancing AI research and investment. Over the past year, DeepSeek has released several competitive AI models, including its V3 model, which raised concerns over content restrictions on sensitive topics related to the Chinese government. However, the release of R1 has pushed the company into the global spotlight.

R1, unveiled last week, has shocked the industry with its low-cost operation and open-source framework, allowing other companies to build on the model. Within days, the DeepSeek app surpassed ChatGPT on app store charts, with nearly two million downloads.
AI is traditionally a cost-intensive technology, requiring massive investments in electricity, data centers, and high-power chips. For example, Meta announced plans to spend $65 billion on AI development this year, while OpenAI CEO Sam Altman has projected the industry will need trillions of dollars in investments. DeepSeek’s ability to achieve comparable results with significantly fewer resources challenges these assumptions and shifts the narrative on AI costs.
Marc Andreessen, co-founder of Andreessen Horowitz, called R1 “one of the most amazing and impressive breakthroughs I’ve ever seen,” underscoring the transformative potential of this achievement. If such advancements can be achieved at lower costs, it may pave the way for broader accessibility to AI technology while raising questions about industry spending and efficiency.
The United States has long relied on export restrictions to maintain dominance in critical technologies. However, DeepSeek’s success raises doubts about the effectiveness of such strategies. Despite sanctions aimed at restricting China’s access to advanced AI chips, DeepSeek has delivered a model that narrows the gap between U.S. and Chinese AI capabilities.
Wall Street’s response was immediate and negative, with Nvidia (NVDA) stocks dropping 12% in premarket trading and significant losses for other tech giants like Meta, Alphabet, and Oracle. The development has heightened concerns about the sustainability of U.S. investments in AI and the potential for overspending.
Although DeepSeek’s achievements are impressive, some industry experts caution against overestimating their impact. The company has not disclosed the full cost of training R1, leaving room for speculation about hidden expenses. Additionally, R1’s capabilities are currently focused on consumer-facing applications, with no evidence that it can rival the more ambitious AI projects requiring massive infrastructure investments.
“The DeepSeek model rollout is leading investors to question the lead U.S. companies have and whether the spending will lead to profits,” said Keith Lerner, analyst at Truist. “Ultimately, U.S. companies remain leaders in AI, supported by rich talent and capital.”
Giuseppe Sette, president of AI market research firm Reflexivity, echoed this sentiment, stating, “The U.S. remains the most promising home turf for the emergence of the first self-improving AI.”
Despite skepticism, DeepSeek’s R1 model has undoubtedly shifted the global conversation on AI development, costs, and competition, forcing the industry to rethink its approach to innovation and efficiency.

