DeepSeek: A New Star in the AI Firmament
Artificial intelligence has just acquired a new celebrity: DeepSeek, a venture of the quantitative analysis (quant) company High-Flyer Capital Management, based in Hong Kong. DeepSeek made waves across Silicon Valley and beyond this week with the release of DeepSeek R1, a new open-source large reasoning model. Impressively, this model rivals OpenAI’s o1, one of their most formidable proprietary models, and does so at a considerably lower cost.
The appearance of DeepSeek R1 has shaken up the already tumultuous, competitive AI market. Previously, OpenAI, Anthropic, and Google were locked in a fierce rivalry to develop the most potent proprietary models, while Meta Platforms frequently intervened with comparable, open-source options.
This time, however, the game-changer is that the company behind this remarkable model is situated in China, a geopolitical rival of the United States. It’s worth noting that until recently, China’s tech sector was widely perceived as inferior to Silicon Valley’s.
Existential Angst in the West
DeepSeek’s rapid ascent has understandably caused unease and soul-searching among tech experts in the United States and the Western bloc. The dominance of OpenAI and the underlying big-tech tactic of funneling more funds and computational resources (such as GPUs) into developing increasingly powerful models are now being questioned.
A Positive Reception from Some Western Tech Leaders
Despite the unease, some Western technology executives have publicly welcomed DeepSeek’s rapid rise.
Marc Andreessen, co-creator of the pioneering Mosaic web browser, co-founder of Netscape, and current general partner at the renowned Andreessen Horowitz venture capital firm, took to Twitter to express his awe: “Deepseek R1 is one of the most amazing and impressive breakthroughs I’ve ever seen — and as open source, a profound gift to the world [robot emoji, salute emoji].”
Yann LeCun, Meta’s Chief AI Scientist, also took to social media to share his reflections:
“To people who see the performance of DeepSeek and think: ‘China is surpassing the US in AI.’ You are reading this wrong. The correct reading is: ‘Open source models are surpassing proprietary ones.’DeepSeek has profited from open research and open source (e.g. PyTorch and Llama from Meta) They came up with new ideas and built them on top of other people’s work. Because their work is published and open source, everyone can profit from it. That is the power of open research and open source.”
Meta’s Response: A Commitment to Open Source
In response to DeepSeek’s ascent, Mark Zuckerberg, founder and CEO of Meta AI, took to Facebook to assure the public of the company’s commitment to open source AI. He promised that a new version of Meta’s open-source Llama AI model, known as Llama 4, would be “the leading state of the art model” when it is released later this year.
Zuckerberg outlined Meta’s ambitious plans for the future of AI, including the construction of a massive 2 gigawatt datacenter, the deployment of over 1 gigawatt of compute in 2025, and the acquisition of over 1.3 million GPUs by the end of the year.
The Future of AI: A Multiplicity of Models?
While Zuckerberg emphasizes open-source AI, it’s clear that he doesn’t fully embrace DeepSeek’s strategy of favoring efficiency over the massive GPU investments of other major labs. The question remains: which vision of the future will dominate the AI landscape? Will one provider emerge as the undisputed leader, or will we witness a fragmented market with multiple models sharing the market?
The competition is intensifying, and only time will tell the outcome of this fascinating race.