Since the release of DeepSeek’s open version of its reasoning model R1 earlier this week, the tech industry has been abuzz with strong opinions about the company’s achievement and its implications for the future of AI.
Praise for DeepSeek’s Breakthrough
Venture capitalist Marc Andreessen described DeepSeek as “one of the most amazing and impressive breakthroughs I’ve ever seen.”
R1 appears to match or outperform OpenAI’s o1 model in certain AI benchmarks. Moreover, the company claims it only cost $5.6 million to train its model, a stark contrast to the hundreds of millions that leading American companies spend on training their own models.
DeepSeek also seems to have achieved this despite U.S. sanctions that restrict the sale of advanced chips to Chinese companies. MIT Technology Review suggests these sanctions are driving startups like DeepSeek to innovate by focusing on efficiency, resource pooling, and collaboration. On the other hand, the Wall Street Journal reported that DeepSeek’s Liang Wenfeng recently told China’s premier that American export restrictions remain a significant bottleneck.
Curai CEO Neal Khosla presented a more skeptical view, claiming that DeepSeek is merely a “ccp state psyop” that is “faking the cost” in an attempt to damage the competitiveness of U.S. AI companies. (A Community Note attached to Khosla’s post pointed out that he offers no evidence for his claim, and that his father, Vinod, is an investor in OpenAI.)
DeepSeek’s Potential Impact on U.S. Markets
Journalist Holger Zschaepitz argued that DeepSeek could pose the “biggest threat to US equity markets.” He pointed out that if a Chinese company can build an advanced model at such a low cost without access to advanced chips, it would raise questions about the value of the hundreds of billions being invested in the AI sector by U.S. companies.
In response, Y Combinator CEO Garry Tan suggested DeepSeek’s success could actually benefit American competitors. He argued that if training models become cheaper and easier, the demand for AI in real-world applications would increase, ultimately driving the demand for computing power and benefiting all players in the ecosystem.
Meta’s Chief AI Scientist Yann LeCun took a different stance, advising against framing the debate as a U.S. vs. China issue. He argued that the real takeaway is that “open source models are surpassing proprietary ones.”
“DeepSeek has profited from open research and open source (e.g., PyTorch and Llama from Meta),” LeCun wrote on LinkedIn this week. “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.”
The debate appears to be driving consumer interest in DeepSeek’s product. As of Sunday afternoon, DeepSeek’s AI assistant has become the top free app in the Apple App Store, surpassing even ChatGPT.
What The Author Thinks
DeepSeek’s rise highlights an essential shift in AI development—open-source models are driving innovation in ways that proprietary models cannot match. The fact that DeepSeek has built on open research and open-source tools like PyTorch and Llama shows that the collaborative nature of open-source development can foster faster and more cost-effective progress in AI. Rather than focusing on geopolitical rivalries, the focus should be on how these open innovations push the entire industry forward. The future of AI is not about hoarding resources, but about building on the work of others to create something better for everyone.
Featured image credit: Freepik
Follow us for more breaking news on DMR