The unveiling of a more cost-effective artificial intelligence model by China’s DeepSeek has rattled the U.S. stock market this week. The repercussions were particularly felt during the week with significant losses for Nvidia, the leading Silicon Valley company renowned for its expensive semiconductor technology that fuels the AI revolution. The global investors dumped tech stocks on Monday Jan. 27 as they worried that the emergence of a low-cost Chinese artificial intelligence model would threaten the dominance of AI leaders evaporating $593 billion of the chipmaker’s market value, a record one-day loss for any company on Wall Street.
Last week, Chinese startup DeepSeek launched a free AI assistant which uses less data at a fraction of the cost of incumbent services. Within a week, the assistant had overtaken U.S. rival ChatGPT in downloads from Apple’s app store. This led the tech-heavy Nasdaq to fall 3.1% on Monday. Nvidia experienced the most significant decline on the Nasdaq, with its shares plummeting nearly 17%. This drop marked a historic one-day loss in market capitalization for any stock on Wall Street.
During the week, U.S. AI-related stocks had rallied sharply after President Donald Trump announced a private-sector plan for what he said would be a $500 billion investment in AI infrastructure through a joint venture known as “Stargate.”
But why the Silicon Valley executives showering praise on DeepSeek-V3 and DeepSeek-R1. The Hangzhou-based startup behind DeepSeek remains largely enigmatic, with limited publicly available information. According to records, its controlling shareholder is Liang Wenfeng, who also co-founded the quantitative hedge fund High-Flyer. Its researchers wrote in a paper last month that DeepSeek-V3 model, launched on Jan. 10, used Nvidia’s lower-capability H800 chips for training, spending less than $6 million. The Hangzhou-based startup behind DeepSeek remains largely enigmatic, with limited publicly available information. According to records, its controlling shareholder is Liang Wenfeng, who also co-founded the quantitative hedge fund High-Flyer. DeepSeek R1 is one of the most amazing and impressive breakthroughs and as open source, a reflective new year’s present for the world.
DeepSeek looks like a ChatGPT-like AI model called R1. It has many similar and conversant capabilities but operating at a fraction of the cost of OpenAI’s, Google’s or Meta’s or Google’s prevalent AI models. They had spent just $5.6 million on computing power for its base model as compared to billions of dollars the US companies spend on their AI technologies. This spectacular and remarkable achievement from a relatively unknown AI startup becomes even more shocking when considering that the USA for years has worked to restrict the supply of high-power AI chips to China, citing national security concerns. Just a week before leaving office, the Biden administration has issued new restrictions on the export of US-developed computer chips that power artificial intelligence (AI) systems, in a final effort to prevent rivals like China from accessing the advanced technology. These curbs, the culmination of years of attempts to block China from gaining ground in its military and industrial leadership efforts, are expected to further inflame tensions between Washington and Beijing. This triggered intense criticism from US tech giants like Nvidia and Oracle.
DeepSeek, an innovative AI development firm situated in Hangzhou, China, was founded in May 2023 by the visionary Liang Wenfeng, a proud alumnus of Zhejiang University. In addition to his role at DeepSeek, Wenfeng is also the co-founder of High-Flyer, a prominent quantitative hedge fund in China that owns DeepSeek. Although DeepSeek operates as an independent AI research lab under the High-Flyer umbrella, the details regarding its funding and valuation remain undisclosed to the public, shrouding the company’s financial backing in mystery.
DeepSeek is dedicated to the development of open-source large language models (LLMs). The company’s inaugural model was launched in November 2023, and since then, they have continuously refined and expanded upon their core LLM, producing several distinct variations. However, it was not until January 2025, with the introduction of their groundbreaking R1 reasoning model, that DeepSeek gained global recognition and fame. DeepSeek represents the latest challenge to OpenAI, which established itself as an industry leader with the debut of ChatGPT in 2022. OpenAI has helped push the generative AI industry forward with its GPT family of models, as well as its o1 class of reasoning models. While the two companies are both developing generative AI LLMs, they have different approaches.
DeepSeek uses a different approach to train its R1 models than what is used by OpenAI. The training involved less time, fewer AI accelerators and less cost to develop. DeepSeek’s aim is to achieve artificial general intelligence, and the company’s advancements in reasoning capabilities represent significant progress in AI development. Since the company was created in 2023, DeepSeek has released a series of generative AI models. With each new generation, the company has worked to advance both the capabilities and performance of its models.
On Jan. 27, 2025, DeepSeek reported large-scale malicious attacks on its services, forcing the company to temporarily limit new user registrations. The timing of the attack coincided with DeepSeek’s AI assistant app overtaking ChatGPT as the top downloaded app on the Apple App Store. Despite the attack, DeepSeek maintained service for existing users. The issue extended into Jan. 28, when the company reported it had identified the issue and deployed a fix.
Memory chips play a crucial role in training large language models like ChatGPT. In the context of neural network training, memory chips, specifically DRAM (Dynamic Random Access Memory), are used to store the model parameters, intermediate activations, and gradients during the training process. These parameters and activations are large arrays of floating-point numbers representing the neural network’s weights and activations at different layers. NVIDIA’s focus on GPU technology, in contrast to the prevailing CPU technology, was a key factor in establishing both its market differentiation and competitive advantages. Whether by design or inherent capacity and improvement, GPU technology excels at parallel processing—utilizing more cores in order to allow multiple processes to run simultaneously. In comparison, CPUs utilize serial processing, which generally requires the completion of one process before the next one can begin.
Although the cost-saving achievement may be significant, the R1 model is a ChatGPT competitor – a consumer-focused large-language model. It hasn’t yet proven it can handle some of the massively ambitious AI capabilities for industries that – for now – still require tremendous infrastructure investments.
The bottom line is the US outperformance has been driven by tech and the lead that US companies have in AI. However, the DeepSeek model rollout is leading investors to question the lead that US companies have and how much is being spent and whether that spending will lead to profits or overspending.
The author, Nazir Ahmed Shaikh, is a freelance writer, columnist, blogger, and motivational speaker. He writes articles on diversified topics. He can be reached at nazir_shaikh86@hotmail.com