DeepSeek, a relatively new player in the AI industry, made headlines on Monday with bold claims about its AI models offering a cost-effective alternative to large language models (LLMs) such as those developed by OpenAI. The announcement resulted in a significant impact on the market, with shares of leading AI chip suppliers NVIDIA and Broadcom plummeting by 17% on the same day. NVIDIA experienced a market capitalization loss of $589 million, setting a new single-day record.
The startup introduced its AI models – including DeepSeek-V3, DeepSeek-R1, and Janus-Pro – which it asserts can rival the performance of models from established AI competitors like OpenAI and Stability AI at a fraction of the cost and chip requirements. The claim was supported by a paper submitted to the arXiv pre-print service last week, showcasing the capabilities of the DeepSeek-R1 reasoning model when compared to OpenAI’s model.
Despite the unverified nature of these claims, both OpenAI and NVIDIA welcomed DeepSeek as a new competitor in the AI market. OpenAI CEO Sam Altman praised the company for its innovative models and contribution to the field of artificial intelligence. A spokesperson from NVIDIA also acknowledged DeepSeek as an “excellent AI advancement” that exemplifies how companies can develop new AI models using existing resources efficiently.
If DeepSeek’s claims hold true, the introduction of low-cost AI models could have significant implications for the industry. Not only could it lead to wider adoption of AI tools, particularly in the realm of cybersecurity where defense against cyber threats is crucial, but it could also help address the power consumption concerns associated with AI data centers.
However, the excitement surrounding DeepSeek’s breakthrough was quickly overshadowed by reports of “large-scale malicious attacks” targeting the company’s services. This development forced DeepSeek to temporarily halt signups, citing security concerns and a need to address the ongoing attacks. The service status page indicated that efforts were being made to resolve the issue and limit registrations to ensure continuous service.
The nature of these attacks remains unspecified, leaving room for speculation about their origin and intent. Some observers suggested the possibility of hacktivist involvement or a surge in suspicious sign-ups overwhelming the company’s infrastructure. Additionally, an AI security researcher reported being able to exploit security flaws in DeepSeek with relative ease, raising further concerns about the company’s vulnerability to cyber threats.
As the situation unfolds, the AI community waits for more information to emerge about DeepSeek and the circumstances surrounding the alleged attacks. Despite the challenges faced by the company, its disruptive impact on the AI market is undeniable, marking a significant moment in the evolution of artificial intelligence in 2025.