Why US AI firms are uneasy with China's Deepseek

In the past year, ChatGPT developed by OpenAI has had great success in the Artificial intelligence (AI) space but that has come at a cost of over $5 billion (Sh645 billion). Now, Deepseek, a Chinese startup has launched its AI model that is much cheaper and has disrupted Silicon Valley.

The AI startup has generated a lot of buzz in recent weeks as a fast-growing rival to OpenAI's ChatGPT, Google's Gemini and other leading AI tools.

So much that NVIDIA, the Chip maker who supplied the H800 chips to Deepseek has lost 17 per cent of its stock accounting for just under 600 billion dollars.

Powered by DeepSeek-V3 which cost less than $6 million (Sh774 million), the AI Chatbot is now top on the US app store, beating ChatGPT.

This casts a dark shadow on the expensive infrastructure the US has committed towards AI infrastructure.

Why DeepSeek is causing a stir

Deepseek is able to offer as much quality cost-effectively raising eyebrows on what the other Chatbots have made us believe. The bullish Chinese startup believes DeepSeek-V3 and DeepSeek-R1, are on par with OpenAI and Meta's most advanced models.

One thing about Silicon Valley and the US is that they made everyone believe Chatbots could be another expensive, exclusive tech that everyone will have to outsource, but the cost of this model is 20 to 50 times cheaper making this Chatbot endeavour an open market.

In 2020 when ChatGPT launched, Chinese tech companies scrambled to react hitting misses, in fact, the release of a ChatGPT equivalent made by Baidu was a disappointment but finally we have DeepSeek.

Scale AI CEO Alexandr Wang said during an interview with CNBC on Thursday, without providing evidence, that DeepSeek has 50,000 Nvidia H100 chips, which he claimed would not be disclosed because that would violate Washington's export controls that ban such advanced AI chips from being sold to Chinese companies.

DeepSeek did not immediately respond to a request for comment on the allegation.

Bernstein analysts on Monday highlighted in a research note that DeepSeek's total training costs for its V3 model were unknown but were much higher than the $5.58 million (Sh720 million) the startup said was used for computing power. The analysts also said the training costs of the equally-acclaimed R1 model were not disclosed.

Liang Wenfeng is the majority shareholder of the Chinese startup whose mark in the AI sector has raised questions about whether the pricing of these models is fair and is the AI bubble about to burst.