DeepSeek's launch of its R1 AI model has turned out to be nothing short of a gigantic seismic event for the tech spheres of the US and the EU, long lulled into complacency by the stringent US export controls on China's ability to procure advanced chips. Now, it seems that China is pressing its strategic advantage by pledging massive funding for its still-nascent AI industry.
Deepseek R1 is AI's Sputnik moment.
— Marc Andreessen 🇺🇸 (@pmarca) January 26, 2025
DeepSeek's launch of its open-source R1 model is now being dubbed "AI's Sputnik moment." For the benefit of those who might not be aware, the model matches - and often outperforms - OpenAI's cutting-edge o1 AI model.
7/ The results are mind-blowing:
- Training cost: $100M → $5M
- GPUs needed: 100,000 → 2,000
- API costs: 95% cheaper
- Can run on gaming GPUs instead of data center hardware— Morgan Brown (@morganb) January 27, 2025
In what is the real shocker, however, DeepSeek's solution is extremely efficient, entailing around 1/50th the training cost of the comparable offerings from its Western counterparts. Combine this with R1's open-source credentials, and you have an instant winner.
this guy makes the FIRST technical argument in this Deep Copendemonium. Kudos!
> You're doing everyone a disservice; it doesn't matter if its 50k H100s, or 30k, or 10k. Whatever it is, it is certainly NOT solely 2k H800s for $5m of training
perhaps.
My math yields 22.4% MFU tho https://t.co/HD3iNLFxbA pic.twitter.com/TeZexTHyby— Teortaxes▶️ (@teortaxesTex) January 25, 2025
Of course, DeepSeek claims that its R1 model was trained on just 2000 H800 GPUs. As the analysis in the above X post shows, the training cluster was likely a lot bigger than a mere 2000 H800s. But the underlying arguments about DeepSeek's cost efficiency remain. After all, the company is charging just 3 percent of what OpenAI typically charges for its premium service.
China Pledges 1 Trillion Yuan In Funding For Its AI Industry
This brings us to the crux of the matter. As the tech favorites in the US tumble on DeepSeek-induced jitters, with NVIDIA alone down 13 percent in early pre-market trading, China has pledged 1 trillion yuan in new funding to develop its AI industry:
"The Bank of China plans to provide special comprehensive financial support with a total scale of no less than 1 trillion yuan to various entities in the entire artificial intelligence industry chain in the next five years ..."
The funding aims to bolster China's self-reliance in the AI sphere, improve AI-related infrastructure (read data centers), and "boost innovation."
Of course, this comes as the US has announced its own Stargate project, which envisages at least $500 billion invested in the US over the next four years to build out the requisite AI infrastructure.
Do you think DeepSeek's R1 model represents a seminal shift in the AI industry? Let us know your thoughts in the comments section below.
