The emergence of DeepSeek is a "huge opportunity" for them, not a threat.
Open source means that the source code of software can be made freely available on the web for modification and redistribution. Unlike competitors such as OpenAI, DeepSeek's model is open source.
R1 shows that growth will not be dominated by one company - there is no hardware and software 'moat' in the open source model.
DeepSeek's open source R1 inference model, released late last month, is comparable to the best technology in the United States and has achieved cutting-edge performance at low cost, shocking the global market.
Just like in the PC and Internet markets, falling prices are helping to drive global adoption. The AI market is on a similar long-term growth path.
Inference chip 'burst'
By accelerating the AI cycle from the "training phase" to the "inference phase," DeepSeek is likely to increase the adoption of new inference chip technology.
Inference refers to the act of using AI to make predictions or decisions based on new information, rather than the "training phase" of building or training models.
Ai training is about building a tool or algorithm, and reasoning is about putting that tool in practical use."
Ai training requires a lot of computation, but reasoning can be done with less powerful chips that are programmed to perform a narrow range of tasks.
Many in the industry believe that as customers adopt and build DeepSeek's open source model, they see increasing demand for inference chips and computing.
DeepSeek has demonstrated that smaller open source models can be trained to be as powerful, if not more powerful, than large proprietary models at a fraction of the cost. With the widespread use of small functional models, they catalyzed an era of reasoning.
He also noted that the company has recently seen a doubling of interest from customers around the world in accelerating its inference initiatives.
Now companies are shifting spending from training clusters to reasoning clusters. We just need more and more computing power to scale these models for millions of users.
As the overall demand for AI grows, smaller companies will have more room to grow, and since the world will need more tokens (units of data processed by AI models), Nvidia won't be able to supply enough chips for everyone, so that gives us the opportunity to sell more aggressively to the market.
