# DeepSeek-R1 Release and Market Shock (January 2025)

DeepSeek, a Chinese AI company, released its open-weight R1 model in January 2025, triggering a major stock market selloff in tech and prompting industry reassessment of AI cost and efficiency.

DeepSeek, a Chinese artificial intelligence company based in Hangzhou, released its open-weight large language model DeepSeek-R1 in January 2025. The release triggered a significant selloff in global technology stocks, as investors reassessed the competitive landscape of AI development and the necessity of massive capital expenditures. The event prompted widespread industry discussion about the efficiency of AI training and the viability of open-source models.

DeepSeek is a subsidiary of High-Flyer, a Chinese hedge fund co-founded by Liang Wenfeng. The company focuses on developing open-weight models, meaning the model parameters are publicly shared, though training data is not openly licensed. Since January 2025, DeepSeek has released its models under free and open-source software licenses. The company's approach emphasizes research over immediate commercialization, allowing it to navigate certain regulatory provisions.

## Background

DeepSeek was founded in July 2023, following the establishment of an artificial general intelligence (AGI) research lab by High-Flyer in April 2023. The lab was spun off into an independent company two months later, with High-Flyer as its principal investor. Liang Wenfeng, who had been trading since the 2008 financial crisis, co-founded High-Flyer in June 2015. The hedge fund began using GPU-dependent deep learning models for stock trading in October 2016, transitioning from CPU-based linear models.

DeepSeek's origins are rooted in High-Flyer's computing infrastructure. In 2019, the company constructed its first computing cluster, Fire-Flyer, at a cost of 200 million yuan. The cluster contained 1,100 GPUs interconnected at 200 Gbit/s and was retired after 1.5 years. By 2021, Liang had begun acquiring large quantities of Nvidia GPUs, reportedly obtaining 10,000 Nvidia A100 GPUs before U.S. restrictions on chip sales to China took effect.

## Fire-Flyer 2

Construction of Fire-Flyer 2 began in 2021 with a budget of 1 billion yuan. By 2022, its capacity was used at over 96%, totaling 56.74 million GPU hours. The cluster had 5,000 PCIe A100 GPUs in 625 nodes, each containing 8 GPUs. Initially, it used PCIe instead of the DGX version of A100 because the models trained could fit within a single 40 GB GPU VRAM, requiring only data parallelism. Later, it incorporated NVLinks and Nvidia Collective Communications Library (NCCL) to train larger models requiring model parallelism.

Fire-Flyer 2 remains in operation as of 2025. It features a co-designed software and hardware architecture. The hardware uses Nvidia GPUs with 200 Gbps interconnects, divided into two zones with two fat trees for high bisection bandwidth. The software includes 3FS (Fire-Flyer File System), a distributed parallel file system designed for asynchronous random reads, using Direct I/O and RDMA Read. It also includes hfreduce, a library for asynchronous communication.

## DeepSeek-R1 Release

DeepSeek-R1 was launched in January 2025 alongside an eponymous chatbot. The model demonstrated competitive performance in reasoning tasks, reportedly rivaling leading models from OpenAI and other Western AI companies. Its open-weight nature allowed developers and researchers to inspect and modify the model, fostering rapid adoption and experimentation.

The release had immediate financial repercussions. On the day of the announcement, major technology stocks experienced sharp declines. Nvidia, a leading GPU manufacturer, saw its market value drop significantly, as investors questioned whether the high demand for AI chips would persist if models could be trained more efficiently. Other companies in the AI supply chain, including [AMD](https://www.wikiprompt.org/wiki/amd), [TSMC](https://www.wikiprompt.org/wiki/tsmc), and [Broadcom](https://www.wikiprompt.org/wiki/broadcom), also faced selloffs. The market shock was attributed to DeepSeek's ability to achieve high performance with relatively lower computational costs, challenging the assumption that massive compute investments were necessary for cutting-edge AI.

## Market Impact and Industry Reassessment

The DeepSeek-R1 release prompted a broad reassessment of AI strategies across the industry. Analysts and executives began to question the sustainability of the enormous capital expenditures by companies like [OpenAI](https://www.wikiprompt.org/wiki/openai), [Anthropic](https://www.wikiprompt.org/wiki/anthropic), and [Google DeepMind](https://www.wikiprompt.org/wiki/google-deepmind). The event highlighted the potential of open-source models to disrupt proprietary offerings, as DeepSeek's model was freely available, reducing barriers to entry for smaller players.

In response, some companies accelerated their own efficiency efforts, focusing on algorithmic improvements and hardware optimization. The incident also intensified discussions about U.S. export controls on advanced chips, as DeepSeek's success demonstrated that Chinese companies could achieve competitive results despite restrictions. The market shock was seen as a wake-up call for investors, who had previously assumed that AI leadership was tied to access to the most advanced hardware.

## Company Operations and Strategy

DeepSeek is headquartered in Hangzhou, Zhejiang, and is owned and funded by High-Flyer. Liang Wenfeng serves as CEO, holding an 84% stake through two shell corporations as of May 2024. The company focuses on research and has stated it has no immediate plans for commercialization. This posture allows it to skirt certain provisions of China's AI regulations aimed at consumer-facing technologies.

DeepSeek's hiring approach emphasizes skills over lengthy work experience, resulting in many hires fresh out of university. The company also recruits individuals without computer science backgrounds to expand the range of expertise incorporated into the models, such as poetry or advanced mathematics. According to The New York Times, dozens of DeepSeek researchers have or have previously had affiliations with People's Liberation Army laboratories and the Seven Sons of National Defence. Since 2025, its chatbot has been adopted in non-combat roles by the People's Liberation Army, including hospitals, the People's Armed Police, and national mobilization organizations.

Due to U.S. chip restrictions, DeepSeek has refined its algorithms to maximize computational efficiency, leveraging older hardware and reducing energy consumption. The company has also expanded into Africa, offering more affordable and less power-hungry AI solutions. DeepSeek has bolstered African language models and generated startups, for example in Nairobi. Along with Huawei's storage and cloud computing services, the impact on the tech scene in sub-Saharan Africa is considerable, offering local data sovereignty and flexibility compared to Western AI platforms.

## Cloud Hosting and Adoption

Due to its open-weight framework, DeepSeek models have been hosted natively by cloud providers including [Microsoft Azure](https://www.wikiprompt.org/wiki/azure) and Perplexity AI. This has facilitated widespread access to the models, allowing developers to integrate them into various applications. The open-weight approach has also enabled third-party fine-tuning and customization, contributing to the model's rapid adoption.

The release of DeepSeek-R1 has been compared to historical moments in technology where disruptive innovations reshaped markets. It underscored the potential for non-U.S. companies to lead in AI development, challenging the dominance of American tech giants. The event also raised questions about the long-term viability of closed-source AI models, as open-source alternatives continue to improve.

## Future Outlook

Following the market shock, DeepSeek continued to develop and release new models. In February 2026, Anthropic accused DeepSeek of using thousands of fraudulent accounts to generate millions of conversations with Claude to train its own LLMs. In April 2026, investors began discussions with DeepSeek for a $300 million funding round, which would bring the company to a total valuation of $10 billion. DeepSeek completed a May 2026 Series A round receiving US$7 billion, reaching a post-money valuation of US$52 billion. In July 2026, Bloomberg and the Financial Times reported that the company had begun preparations for an IPO, potentially listing as soon as 2027. The same month, it began discussions targeting a US$70 billion pre-money valuation.

The January 2025 release of DeepSeek-R1 remains a landmark event in the history of [artificial intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence), illustrating the rapid pace of innovation and the interconnectedness of technology and financial markets. The event prompted a global conversation about AI efficiency, open-source development, and the geopolitical dimensions of AI advancement.

---
Source: https://www.wikiprompt.org/wiki/deepseek-r1-release-and-market-shock-jan-2025
License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/)
Last updated: 2026-09-13T03:52:15.813865+00:00