DeepSeek, a relatively unknown company in China until just two years ago, has taken the world by surprise with its innovative AI tool, threatening to disrupt the tech economy. On Monday, DeepSeek caused a significant upheaval in the US stock markets, wiping out about $1 trillion, marking the biggest jolt in America’s share trading history.
Overnight, its smartphone app became the most downloaded on the Apple store in America. Despite this rapid rise to fame, several basic questions about DeepSeek and its AI tool remain. Here, we aim to answer some of these questions.
What is DeepSeek?
DeepSeek is an artificial intelligence company based in China. Its AI application can be accessed as an Android app, iOS app, via a web browser, or through APIs (Application Programming Interfaces) for integration within other apps.
What Can Users Do with DeepSeek?
DeepSeek’s functionalities are similar to those of other AI tools like OpenAI’s ChatGPT, Google Gemini, and Microsoft Copilot. Users can generate answers to questions, assist in research, aid in writing, editing, and proofreading, translate content between different languages, extract and analyze text from files, summarize documents, and help in writing and debugging software code, among other tasks.
Who Owns DeepSeek?
DeepSeek is owned and funded by the Chinese hedge fund company High-Flyer. Liang Wenfeng, the co-founder of High-Flyer, established DeepSeek in 2023 and currently serves as its CEO.
Is DeepSeek New?
Not exactly. The company was founded in May 2023 and released its first model, DeepSeek Coder, in November 2023.
What Makes DeepSeek Different from Other AI Models?
DeepSeek claims to have significantly reduced the compute and data resources needed for its operations, training its AI at a fraction of the cost of OpenAI. This cost efficiency is also reflected in its API costs, which are more than 90% cheaper than comparable models from OpenAI. Additionally, DeepSeek’s models are open source, allowing anyone to use the key components to build their own tools for free. While Meta’s Llama is also open source, OpenAI’s ChatGPT is not.
Why the Hullabaloo Now?
On January 20, DeepSeek released its new large language model (LLM) DeepSeek-R1, which performed as well as or better than OpenAI’s model in performance tests. This achievement, coupled with the claim that DeepSeek achieved this at a fraction of the cost, captured the interest of the technology and financial communities.
Silicon Valley capitalist Marc Andreessen described it as “AI’s Sputnik moment,” drawing a parallel with the launch of the Sputnik 1 satellite by the Soviet Union in 1957, which caught the US unawares and triggered the space race. This led to questions about the scale of American investments in AI and a market reaction that shaved off almost $600 billion from chip-maker Nvidia’s market cap, the biggest single-day loss in US history. The buzz around DeepSeek made its AI assistant the No. 1 downloaded free app on Apple’s iPhone store in the US.
Why Nvidia?
Nvidia is an American computer chip design company that manufactures its products in Taiwan. Nvidia specializes in graphics processing units (GPUs), which enhance image and video rendering on computers. GPUs can perform calculations concurrently in a way that regular central processing units (CPUs) cannot. This capability is highly valuable for AI companies, which rely on a high volume of computations to build their LLMs. Nvidia’s first-mover advantage and existing supply chain enabled it to better meet the needs of emerging AI companies.
Nvidia shares have surged with the AI boom, and on October 26, 2024, Nvidia even overtook Apple as the world’s most valuable company, with its market value touching $3.53 trillion.
But Didn’t the US Restrict Exports of AI Chips to China? How Did DeepSeek Build Its AI?
Yes, the US has restricted exports of AI chips to China. However, DeepSeek’s parent company High-Flyer acquired over 10,000 Nvidia A100 GPUs in 2022 before the export restrictions were imposed. DeepSeek also stated that its recent model was built using Nvidia’s lower-performing H800 chip, which is not restricted for sale to China. This has sparked debate about whether top-of-the-line hardware is necessary for building top-notch AI models.
What is LLM?
LLM, or large language model, is a type of artificial intelligence computer program that can recognize and generate text and perform other tasks. These programs are trained on vast amounts of data, generally scraped from the internet.
Do We Know What Data DeepSeek Was Trained On?
While DeepSeek has open-sourced its R1 code under an MIT license, it has not revealed the source of its training data. There have been allegations that DeepSeek’s LLM could have been trained on synthetic data (data produced by other LLMs) instead of original data collected from online sources.
Are There Data Security Concerns with DeepSeek?
According to DeepSeek’s privacy policy, it collects personal data such as text, audio input, and chat history, and this data is stored in China, subjecting it to China’s data-sharing laws.
What About Censorship in China? Does It Extend to DeepSeek’s Responses?
As DeepSeek operates out of China, it must comply with Chinese government regulations, which include censoring topics deemed contentious by Chinese authorities. It deflects any such questions.
What Does DeepSeek Mean for the Larger AI Industry?
As DeepSeek’s code is open source, others can learn from and improve upon it. This is likely to benefit the larger AI community by increasing efficiencies and reducing costs, thereby accelerating AI advancements.
Does DeepSeek Support Indian Languages?
Yes. According to DeepSeek, it supports 22 Indian languages. However, proficiency varies depending on the availability of training data. It claims strong support for 11 Indian languages (Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam, Odia, Punjabi, and Urdu) and moderate to limited support for 11 others (Assamese, Sanskrit, Sindhi, Nepali, Konkani, Maithili, Kashmiri, Manipuri or Meitei, Dogri, Santhali, and Bodo).
ALSO READ: China Withdraws From Global Project To Build World’s Largest Dam