DeepSeek: Everything you need to know about the AI chatbot app
DeepSeek has gone viral.
Chinese AI lab DeepSeek broke into the mainstream consciousness this week after its chatbot app rose to the top of the Apple App Store charts (and Google Play, as well). DeepSeek's AI models, which were trained using compute-efficient techniques, have led Wall Street analysts — and technologists — to question whether the U.S. can maintain its lead in the AI race and whether the demand for AI chips will sustain.
But where did DeepSeek come from, and how did it rise to international fame so quickly?
DeepSeek is backed by High-Flyer Capital Management, a Chinese quantitative hedge fund that uses AI to inform its trading decisions.
AI enthusiast Liang Wenfeng co-founded High-Flyer in 2015. Wenfeng, who reportedly began dabbling in trading while a student at Zhejiang University, launched High-Flyer Capital Management as a hedge fund in 2019 focused on developing and deploying AI algorithms.
In 2023, High-Flyer started DeepSeek as a lab dedicated to researching AI tools separate from its financial business. With High-Flyer as one of its investors, the lab spun off into its own company, also called DeepSeek.
From day one, DeepSeek built its own data center clusters for model training. But like other AI companies in China, DeepSeek has been affected by U.S. export bans on hardware. To train one of its more recent models, the company was forced to use Nvidia H800 chips, a less-powerful version of a chip, the H100, available to U.S. companies.
DeepSeek's technical team is said to skew young. The company reportedly aggressively recruits doctorate AI researchers from top Chinese universities. DeepSeek also hires people without any computer science background to help its tech better understand a wide range of subjects, per The New York Times.
DeepSeek unveiled its first set of models — DeepSeek Coder, DeepSeek LLM, and DeepSeek Chat — in November 2023. But it wasn't until last spring, when the startup released its next-gen DeepSeek-V2 family of models, that the AI industry started to take notice.
DeepSeek-V2, a general-purpose text- and image-analyzing system, performed well in various AI benchmarks — and was far cheaper to run than comparable models at the time. It forced DeepSeek's domestic competition, including ByteDance and Alibaba, to cut the usage prices for some of their models, and make others completely free.
DeepSeek-V3, launched in December 2024, only added to DeepSeek's notoriety.
According to DeepSeek's internal benchmark testing, DeepSeek V3 outperforms both downloadable, openly available models like Meta's Llama and 'closed' models that can only be accessed through an API, like OpenAI's GPT-4o.
Equally impressive is DeepSeek's R1 "reasoning" model. Released in January, DeepSeek claims R1 performs as well as OpenAI's o1 model on key benchmarks.
Being a reasoning model, R1 effectively fact-checks itself, which helps it to avoid some of the pitfalls that normally trip up models. Reasoning models take a little longer — usually seconds to minutes longer — to arrive at solutions compared to a typical non-reasoning model. The upside is that they tend to be more reliable in domains such as physics, science, and math.
There is a downside to R1, DeepSeek V3, and DeepSeek's other models, however. Being Chinese-developed AI, they're subject to benchmarking by China's internet regulator to ensure that its responses "embody core socialist values." In DeepSeek's chatbot app, for example, R1 won't answer questions about Tiananmen Square or Taiwan's autonomy.
If DeepSeek has a business model, it's not clear what that model is, exactly. The company prices its products and services well below market value — and gives others away for free.
The way DeepSeek tells it, efficiency breakthroughs have enabled it to maintain extreme cost competitiveness. Some experts dispute the figures the company has supplied, however.
Whatever the case may be, developers have taken to DeepSeek's models, which aren't open source as the phrase is commonly understood but are available under permissive licenses that allow for commercial use. According to Clem Delangue, the CEO of Hugging Face, one of the platforms hosting DeepSeek's models, developers on Hugging Face have created over 500 'derivative' models of R1 that have racked up 2.5 million downloads combined.
DeepSeek's success against larger and more established rivals has been described as "upending AI" and "over-hyped." The company's success was at least in part responsible for causing Nvidia's stock price to drop by 18% on Monday, and for eliciting a public response from OpenAI CEO Sam Altman.
Microsoft announced that DeepSeek is available on its Azure AI Foundry service, Microsoft's platform that brings together AI services for enterprises under a single banner. When asked about DeepSeek's impact on Meta's AI spending during its first-quarter earnings call, CEO Mark Zuckerberg said spending on AI infrastructure will continue to be a 'strategic advantage' for Meta.
At the same time, some companies are banning DeepSeek, and so are entire countries and governments, including South Korea. New York state also banned DeepSeek from being used on government devices.
As for what DeepSeek's future might hold, it's not clear. Improved models are a given. But the U.S. government appears to be growing wary of what it perceives as harmful foreign influence.
TechCrunch has an AI-focused newsletter! Sign up here to get it in your inbox every Wednesday.
This story was originally published January 28, 2025, and will be updated continuously with more information.
Hashtags

Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Yahoo
23 minutes ago
- Yahoo
U.S., Chinese delegates in London to talk trade, rare earths
June 9 (UPI) -- Delegates from the United States and China are set to meet Monday in London after a phone call between the nations' leaders seemingly led to a cooling of tensions related to their otherwise heated recent trade dispute. "We are a nation that champions free trade and have always been clear that a trade war is in nobody's interests, so we welcome these talks," said a British government spokesperson. The U.K. has provided the space for the countries to chat but hasn't publicly disclosed its location. American attendees are slated to include U.S. Commerce Secretary Howard Lutnick, U.S. Treasury Secretary Scott Bessent and U.S. Trade Representative Jamieson Greer, while Chinese Vice Premier He Lifeng will lead his country's delegation. It is expected the discussion will put a fair amount of focus on the rare earth minerals situation. The Trump administration had expected China to back down on export restrictions it had imposed in April on such minerals after talks held in May. China imposed those restrictions in response to tariffs levied by Trump on Chinese goods. The resulting trade disruption has led to a 2.9% decrease on exports to the United States from April to May, the decrease from May 2024 is 3.4% and the cumulative year-on-year decrease from January to May is at 4.9%, according to Chinese customs data. However President Donald Trump and Chinese President Xi Jinping spoke on the phone last week, and the conversation was reportedly so friendly it not only led to Monday's meeting but each invited the other to make a personal visit. American and Chinese representatives had met last month in Geneva and reportedly reached an agreement to suspend most of the tariffs that had been reciprocally imposed, but both countries have since been accused of agreement violations by the other.
Yahoo
23 minutes ago
- Yahoo
Nvidia (NVDA) Reaffirms AI Dominance After CFO Dinner Briefing
Nvidia (NVDA, Financials) provided new clarity on key investor concerns during a private investor dinner hosted by Bank of America in San Francisco, reinforcing its long-term AI strategy and execution roadmap. Warning! GuruFocus has detected 4 Warning Signs with NVDA. The event featured Nvidia CFO Colette Kress, VP of Investor Relations Toshiya Hari, and VP Ian Buck, who addressed three core concerns: the rollout of the Blackwell rack systems, global AI diffusion including sovereign demand, and the impact of U.S. export restrictions on Chinese AI chip shipments. Blackwell NVL rack systems were said to be fully ramped as of June 5, with roughly 70% of Nvidia's compute sales in fiscal Q1 tied to the platform. The firm also confirmed that production of Blackwell Ultra will begin in the current quarter, with volume expected to rise in the second half. The next-generation Rubin and Feynman architectures remain on schedule, supporting Nvidia's annual product cadence. Bank of America estimated Nvidia's total addressable data center market opportunity between $250 billion and $350 billion, based on third-party projections of 5 to 7 gigawatts of new capacity in 2025. Nvidia's current pipeline stands at roughly $175 billion, supported by what the analyst called a symbiotic relationship with hyperscalers. The analyst added that the company has fully de-risked China in its latest data center forecasts, removing any projected sales tied to the H20 AI chip, which was subject to export licensing restrictions during the first quarter. Nvidia previously disclosed a $2.5 billion sales hit related to those restrictions in its May 28 earnings. The easing of sovereign AI restrictions was also seen as a tailwind, with Nvidia now able to serve government-backed AI demand either directly or through cloud and enterprise intermediaries. Bank of America reiterated a Buy rating and a $180 price objective for Nvidia, citing the company's multi-year lead in performance, pipeline, incumbency, scale, and developer support as key reasons for sustained upside in the AI cycle. Nvidia previously reported fiscal Q1 revenue of $44.1 billion, up 69% year over year. This article first appeared on GuruFocus. Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
Yahoo
25 minutes ago
- Yahoo
TikTok creating more than 500 new British jobs as UK users top 30 million
TikTok creating more than 500 new British jobs as UK users top 30 million TikTok has unveiled plans to create more than 500 new British jobs and open a second office in London as it cheered passing the 30 million milestone for users in the UK. The Chinese-owned social media firm said its UK workforce would expand to 3,000 this year as part of its 'deep commitment' to the UK. It will also open a new 135,000sq ft office in London's Barbican, which is set to open early next year. The group already has its UK headquarters in Farringdon, London, which were opened in 2022. TikTok unveiled the plans as it said it now has more than 30 million regular users in the UK each month, which makes the market is biggest user community in Europe. ADVERTISEMENT Adam Presser, director of TikTok UK and global head of operations and trust and safety, said: 'Whether through direct investment in jobs and innovation, or the wider economic contribution from millions of British businesses on TikTok, we're pleased to be increasing our investment and presence here in the UK, an important hub for TikTok.' But it comes after Cabinet minister Pete Kyle signalled he was looking at measures to restrict the amount of time children spend on their phones, including through a possible 10pm curfew. Mr Kyle was asked on Sunday morning whether he would look at limiting the time children spend on social media to two hours per app after the Sunday People and Mirror reported the measure was being considered by ministers. The Online Safety Act has passed into law, and from this year will require tech platforms to follow new Ofcom-issued codes of practice to keep users safe online, particularly children. Hefty fines and site blockages are among the penalties for those caught breaking the rules, but many critics have argued the approach gives tech firms too much scope to regulate themselves. ADVERTISEMENT TikTok's Mr Presser said that, as well as its UK expansion plans, the group also invests 'significantly' in safety. The firm launched its UK operations in 2018 (James Manning/PA) He said: 'What underpins our continued growth is our deep commitment to safety and to creating an enjoyable and secure digital space to sustainably support creators, entrepreneurs and the wider economy, which is why we also invest significantly in safety.' TikTok first launched its UK operations in 2018 and is financially incorporated in Britain. The group was fined 530 million euro (£446 million) by the Irish data protection watchdog last month for breaching EU privacy rules around transferring user data to China. The video-sharing app was also sanctioned for not being transparent with users about where personal data was being sent and ordered the platform to comply with data protection rules within six months. ADVERTISEMENT TikTok said it would appeal against the decision. The social media giant, which is owned by China-based ByteDance, has been under scrutiny from regulators around the world over how it handles personal data, and is also facing a ban in the United States over its China links, which the US government has said is a national security issue.