A China-based rival to ChatGPT, called DeepSeek, sent shockwaves across the technology industry by reportedly becoming the most downloaded free app on Apple devices and operating with fewer resources than its US counterpart.
DeepSeek unveiled its AI model, DeepSeek-R1, last week, taking on not only OpenAI’s ChatGPT but also Google’s Gemini and Meta’s Llama. Nvidia hailed DeepSeek as “an excellent AI advancement and a perfect example of Test Time Scaling.” Test Time Scaling is a technique that adjusts a large language model during use without requiring a full retraining.
Bloomberg reported that the launch of DeepSeek sparked discussions about the cost-effective use of large language models (LLMs). Unlike many competitors, DeepSeek operates on less powerful chips. This development led to a drop in share prices of companies that have benefited from the AI boom, such as Nvidia. Despite this, UBS analysts maintained their buy rating for Nvidia shares.
UBS estimates that DeepSeek costs just $0.55 per million input tokens and $2.19 per million output tokens, compared to OpenAI’s $15 and $60 for input and output tokens, respectively.
In a statement, UBS said: “If DeepSeek’s model were to prove to be the way forward for the broader AI industry, it is not necessarily a zero-sum game. The overall market can grow, with potentially lower costs accelerating AI adoption across industries and further improving productivity gains.”
The US has imposed restrictions on domestic companies selling software and hardware to Chinese firms in an effort to hinder China’s advancements in AI technology.