Well, alittle thoughts from my own that I would like to pen down about DeepSeek.
DeepSeek has been making waves lately after unveiling its AI app to the world. It’s gained a lot of attention, partly because it launched for free, but also because it’s claiming to outperform OpenAI’s ChatGPT. What sets it apart? DeepSeek says it can handle each prompt at a fraction of the cost, making it more efficient and cost-effective. This is bound to catch the eye of data center operators, government agencies, and enterprises, encouraging them to check out DeepSeek’s models (which are on GitHub—links below) and rethink their current AI setups.
Red Hat is sticking to its core message: offering a platform and toolchains to help businesses run secure AI and machine learning workflows seamlessly across clouds. That said, this doesn’t mean it’s going to completely change how companies approach AI services like OpenAI or Azure AI.
On the front about market sentiments; NVIDIA’s recent stock drop and the broader tech slump, it’s not a sign of a replacement technology or a new AI chip vendor with better total cost of ownership. NVIDIA GPUs are still essential in the long run, and the reliance on them across the ecosystem isn’t going anywhere anytime soon. It’s how the market’s understanding on AI, any news relating to AI will only serve to widen their eyeballs without the need to dive deeper into which parts of AI are we talking about. See the image capture on the news below. Infact it will drive more value for many of the tech stocks (i.e., NVIDIA).
Now, diving alittle into DeepSeek.
DeepSeek is founded by Liang Wenfeng who serves as the CEO, check out wikipedia for more of his background. He seems like a hedge fund business man, but among his team, they purchased 10,000 NVIDIA A100 GPUs prior to US restriction on purchasing AI chips on China. And another source also mentioned his use of 2,000 x H200 GPUs. This is incredible honestly.
It is worth noting that Liang mentioned how when it comes to disruptive technologies, closed source approaches can only temporarily delay others in catching up. This only accelerates the foundation of open source. I subscribe to his beliefs and hope that we’ll hear more about him on the stage along the likes of Jenson Huang / Elon Musk.
Here’s a few background about DeepSeek auto-generated by ChatGPT 4o (paid premium services):-
DeepSeek is a Chinese artificial intelligence company founded in 2023 by Liang Wenfeng, who serves as its CEO. The company is based in Hangzhou, Zhejiang province, and is owned and solely funded by the Chinese hedge fund High-Flyer, which Liang co-founded. Liang Wenfeng, born in 1985 in Guangdong, China, studied electronic information engineering at Zhejiang University. Before establishing DeepSeek, he co-founded High-Flyer, a quantitative hedge fund that leverages artificial intelligence for trading. DeepSeek has gained attention for developing advanced AI models at a fraction of the cost compared to its U.S. counterparts. Its recent model, R1, was developed for just $6 million, contrasting sharply with the hundreds of millions spent by firms like OpenAI. New York Post The company's success has had significant impacts on the global tech industry, including a substantial decline in Nvidia's stock value. theguardian.com Liang Wenfeng's leadership and DeepSeek's innovative approaches have positioned the company as a prominent player in the AI sector, challenging established competitors and contributing to China's advancements in artificial intelligence.
Now the technical portions which I’ve yet to model inference yet. A few links for myself for model inferencing:-
- transformer – https://github.com/huggingface/transformers
- DeepSeek Prover V1.5 – https://github.com/deepseek-ai/DeepSeek-Prover-V1.5?tab=readme-ov-file
Alright that’s all for now. Goodnight.