DeepSeek’s AI app has shaken the AI sector to its core.
Launched last week, the Chinese company’s open-source R1 model is a generative AI (GenAI) chatbot in a similar mold to OpenAI’s ChatGPT platform.
While DeepSeek is still new to the American market, it already appears to be impressing users, with DeepSeek being the most downloaded free app on Apple's US App Store over the weekend.
However, whereas the likes of OpenAI, Google, and Anthropic have been committing extensive resources to their platforms in recent times – with the former having spent $5 billion last year – DeepSeek's R1 model was reportedly built for just $5.6 million.
Unsurprisingly, this news has caused chaos across Wall Street, with American GPU and AI chip designer NVIDIA having a tough week on the stock market.
Indeed, its share price drop equates to $600bn in market value, making it the biggest decrease in the history of the US stock market.
Venture Capitalist Marc Andreessen likened the emergence of a new AI powerhouse to the rivalry between the US and the Soviet Union during the space race, stating:
DeepSeek-R1 is AI's Sputnik moment.
The Future of AI
Andreessen was far from alone in weighing in on the hottest topic in the AI space.
In a post on LinkedIn, Eric Yuan, CEO of Zoom, shared his thoughts on what the arrival of DeepSeek will mean for the future of enterprise AI applications.
Yuan posits that DeepSeek’s tech could actually benefit larger American AI companies, as the Chinese company has shown a way to implement a powerful AI chatbot with lower GPU (graphics processing units) requirements.
The Zoom CEO detailed how AI companies like Anthropic, Meta, OpenAI, and X.AI currently require significant GPU resources to train their models, which leads to high costs.
By adopting DeepSeek's approach, these companies could reduce their GPU needs and lower training expenses.
Judging by the stock market’s response to NVIDIA, many people appear to be operating under the assumption that less reliance on GPUs for the AI giants will negatively impact hardware providers.
However, Yuan argues that it could actually help to balance GPU demand across industries.
