Writer: Harshul Agarwal
Date Written: 04/02/2025
On the 27th of January, 2025, DeepSeek, a Chinese artificial intelligence company, captured
the attention of institutional investors around the world and was responsible for wiping over
USD 1 Trillion off the market. Until now, OpenAI was the leading Large-Language Model
(LLM), available on the market, and now DeepSeek is available on the market, offering its
services for just 5% of the cost of OpenAI. The article will focus on the short-term impacts of
this development on the market, as it is too soon to judge on a long-term basis. We will also
take the time to truly understand DeepSeek, and answer if it’s to stay alongside OpenAI or is
it to remove it from the market.
To understand DeepSeek's initial impacts, it's vital to note its 95% improved cost efficiency
in comparison to OpenAI’s leading LLM model. Due to this fact, Nvidia, whose business
model involves open-selling high-performance GPUs to AI firms running cost-heavy models,
dropped 17% of its market cap, over USD 600 billion. Similarly, firms like SuperMicro
Computer dropped 12.52%, given that they supply cooling solutions for expensive GPUs. But
is DeepSeek as good as everyone and I have made it sound?
On one hand, its ability to solve complex problems in areas such as math, physics, and
reasoning matches or sometimes even surpasses the capabilities of OpenAI’s ChatGPT.
However, when it comes to tasks that require scraping the Internet for information,
DeepSeek's performance was considered inferior to OpenAI’s. Furthermore, concerns arise
about the accuracy of the information provided by DeepSeek, given that its response to
questions about sensitive historical topics can sometimes appear as “Let's Talk about something else.”
Turning back to DeepSeek's broader market impact, many of the American stocks took a
nosedive upon the initial release of DeepSeek, though several recovered the following day
when the market opened, so was it even truly impactful to the market? While the immediate
market reaction was mixed, DeepSeek’s most significant impact was on the mindset of
institutional investors globally. Up until now, OpenAI was the only company offering a truly
effective LLM, making it difficult for investors to compare prices or gauge competition.
Institutions took the weekend evaluating how DeepSeek fairs to OpenAI, and most of them
took the opening day of the market, with two figures, “95% more efficient”, and only “10,000
old generation Nvidia GPUs”. These figures are staggering. However, reports have emerged
that the Chinese government covered the majority of the expenses generated from DeepSeek,
and the fact that DeepSeek has obtained Nvidia GPUs through unofficial channels, however at this point in time this is all speculation that I may say.
We look forward to how DeepSeek's fairs in the longer term in terms of its impact on our
daily lives and the financial markets.
References:
Wiggers, K. (2025). DeepSeek claims its ‘reasoning’ model beats OpenAI’s o1 on certain benchmarks
| TechCrunch. [online] TechCrunch. Available at:
https://techcrunch.com/2025/01/27/deepseek-claims-its-reasoning-model-beats-openais-o1-on-certain-benchmarks/ [Accessed 28 Jan. 2025].
BRACAI (2025). DeepSeek performance vs peers. [online] BRACAI AI Consulting. Available at:
https://www.bracai.eu/post/deepseek-performance [Accessed 28 Jan. 2025].
Cosgrove, E. (2025). DeepSeek’s cost-efficient performance is scaring markets over AI spend. [online]
Business Insider. Available at:
https://www.businessinsider.com/explaining-deepseek-chinese-models-efficiency-scaring-markets-2025-1 [Accessed 28 Jan. 2025].
Comments