What is DeepSeek? Chinese AI startup that has taken over ChatGPT on App Store

In a surprising twist in the AI space, Chinese startup DeepSeek has risen to prominence, challenging some of the biggest players in the industry like OpenAI, Meta, and Google. DeepSeek’s innovative AI models have not only impressed tech experts but also sparked a massive surge in downloads, overtaking ChatGPT on Apple’s App Store in multiple countries. This blog explores what makes DeepSeek so special and why its rise is causing waves in the tech world.

What is DeepSeek?

DeepSeek, founded in 2023 by Liang Wenfeng, is an AI startup that has quickly gained international attention. Liang, who is also the head of the Chinese quant hedge fund High-Flyer, started DeepSeek with the goal of creating an AI company that could rival the likes of OpenAI and Meta while being built at a significantly lower cost. The company’s rise has been rapid, with its app surpassing 1.6 million downloads and topping the App Store charts in several countries, including the US, UK, Australia, Canada, China, and Singapore.

DeepSeek’s Revolutionary Models

What sets DeepSeek apart is its ability to build large language models (LLMs) that perform on par with leading models from US tech giants, but at a fraction of the cost. The company’s flagship models, DeepSeek V3 and DeepSeek R1, have garnered significant attention for outperforming their competitors in various benchmarks, including those from OpenAI and Meta. Despite the lower cost of development, DeepSeek’s models have shown exceptional performance in tasks such as mathematics, coding, and reasoning.

The DeepSeek V3 model has been widely praised for its capability to perform similarly to OpenAI’s latest O1 model. Both R1 and V3 are ranked in the top 10 on the Chatbot Arena, a ranking system affiliated with UC Berkeley, which assesses the performance of leading chatbots. While models like Google’s Gemini 2.0 Flash and ChatGPT 4-o are ranked ahead of DeepSeek, its models outperform those from other competitors like xAI and Anthropic, a company backed by Amazon.

Get Portfolio Insights With Leni Analytics

The Secret Behind DeepSeek’s Cost Efficiency

One of the most striking aspects of DeepSeek’s success is its ability to create cutting-edge AI models at a much lower cost than its competitors. According to DeepSeek’s researchers, the cost of training its V3 model was less than $6 million, a stark contrast to the billions being spent by companies like Microsoft, Meta, and OpenAI. This cost-effective approach has raised eyebrows in the tech world, particularly because AI models traditionally require significant computing power and resources.

The lower cost is partly attributed to the use of Nvidia H800 chips and the clever combination of these with cheaper, less advanced chips. This strategy has allowed DeepSeek to build powerful models without relying on the most expensive hardware, something that has caused a significant shift in how the AI community views the future of AI development.

Why is DeepSeek a Concern for US Companies?

The rise of DeepSeek has raised concerns among US tech companies about their market dominance and pricing strategies. The success of DeepSeek has cast doubt on the prevailing belief that running AI models requires ever-increasing amounts of computing power and expensive hardware. With the company’s ability to develop models cheaply, DeepSeek is putting pressure on other AI companies to reconsider their high development costs. In the same way, in the real estate industry, platforms like Leni challenge the status quo by offering a more efficient way to handle complex data analyticand investment portfolios without exorbitant costs.

Moreover, the export curbs imposed by the Biden administration to slow the growth of AI in countries like China are now facing scrutiny. DeepSeek’s breakthrough suggests that these restrictions may not be as effective as intended, as the company has still managed to achieve remarkable results despite limited access to cutting-edge chips. 

A Global Impact

DeepSeek’s free AI-powered chatbot has caused a storm in the tech industry, particularly when it became the most downloaded app on the Apple App Store in countries such as the US, UK, Australia, and China. The app’s success is a clear indicator of how deeply DeepSeek has resonated with users worldwide. While the chatbot’s performance is often compared to ChatGPT, DeepSeek’s model is uniquely designed to avoid politically sensitive topics, especially those relating to China. For example, when asked about the Tiananmen Square incident, the chatbot refused to comment, responding with, “Sorry, that’s beyond my current scope. Let’s talk about something else.”

While this censorship could limit DeepSeek’s global appeal, it remains a significant player in the AI market, thanks to its cutting-edge technology and competitive pricing.

Silicon Valley’s Reaction to DeepSeek

Silicon Valley has been paying close attention to DeepSeek’s progress. Meta’s Chief AI Scientist, Yann LeCun, acknowledged the company’s success, highlighting how DeepSeek has built its AI on top of open research and open-source frameworks, such as PyTorch and Llama from Meta. 

Venture capitalist Marc Andreessen also expressed his admiration for DeepSeek, calling its R1 model “one of the most amazing and impressive breakthroughs I’ve ever seen,” referring to it as a “profound gift to the world.”

The CEO of Salesforce, Mark Benioff, has praised DeepSeek for its ability to build powerful AI systems without relying on expensive supercomputers. He pointed out that the true value of AI lies in data and metadata, not just the user interface or the model, emphasizing that DeepSeek’s cost-effective approach could redefine the AI industry.

A Boost for China’s AI Ambitions

DeepSeek’s rapid rise is a significant boost to China’s AI ambitions. The success of DeepSeek is seen as a validation of China’s Innovation 2.0 era, where the country is striving to become a global leader in technology, particularly in fields like AI. As China shifts its focus from traditional manufacturing to advanced technologies such as chips, electric vehicles, and AI, DeepSeek represents a key player in this transformation.

China’s government has long prioritized AI development, and DeepSeek’s success is a testament to the growing technological self-reliance within the country. As the global tech industry watches closely, DeepSeek’s rapid ascent could lead to new dynamics in the global AI market, with Chinese companies challenging the established dominance of Silicon Valley giants.

Conclusion

DeepSeek has proven that cutting-edge AI can be developed with lower costs, challenging the status quo held by industry giants like OpenAI, Meta, and Google. With its cost-effective AI models and the success of its free app, DeepSeek is poised to reshape the future of AI development and disrupt the market. As the company continues to gain traction, it is clear that DeepSeek’s rise is a wake-up call for US tech companies, signaling the emergence of a formidable competitor from China in the field of AI.

At Leni, we embrace the evolving landscape of AI and data intelligence, recognizing both the challenges and opportunities it presents. Our commitment to innovation drives us to refine our models, enhance automation, and deliver smarter insights that empower asset managers to optimize their portfolios. As AI reshapes industries, we see this as a catalyst for growth pushing us to stay agile, leverage new advancements, and continue providing cutting-edge solutions that make real estate asset management more efficient and data-driven.

Get started with a free demo of Leni >