At the beginning of the new spring, DeepSeek has become extremely popular in Silicon Valley and Wall Street with its highly 

competitive price advantage and precise algorithms. At the same time, major technology giants have responded rapidly, all 

seamlessly accessing DeepSeek's R1 model service, aiming to seize the development opportunities in the AI field.
Microsoft took the lead. On January 29 local time in the United States, Microsoft announced that DeepSeek R1 had been launched

 on Azure AI Foundry and GitHub. Developers can use this new model to conduct tests and build cloud - based applications and

 services. Microsoft's customers can also run a streamlined version of the DeepSeek R1 model locally on their Copilot+ PCs.
Although Microsoft is an important investor in OpenAI and the two parties have extensive cooperation, in terms of promoting 

product commercialization, Microsoft still tends to adopt a diversified model strategy. Today, the Azure platform not only includes

 OpenAI's GPT series, Meta's Llama series, and Mistral's models, but also DeepSeek has become a new member.Immediately afterwards,

AWS announced that users can deploy the DeepSeek - R1 model on the two major AI service platforms,Amazon Bedrock and Amazon SageMaker AI.

Just one day later, NVIDIA also announced its access to the DeepSeek R1 model. NVIDIA announced on its official website: "To 

help developers safely test these functions and build their own exclusive agents, the 671 - billion - parameter DeepSeek - R1 

model is now available as a preview of the NVIDIA NIM microservice on build.nvidia.com. When running the DeepSeek - R1 NIM 

microservice on a single NVIDIA HGX H200 system, it can process up to 3,872 tokens per second." Developers can use the application

 programming interface (API) for testing and experimentation. This interface is expected to be launched as part of the NVIDIA AI

 Enterprise software platform in the form of a downloadable NIM microservice soon.
Meta's CEO, Mark Zuckerberg, also caught on to the extremely high popularity of DeepSeek. During a conference call with analysts

 on Wednesday, Zuckerberg talked about DeepSeek. He said that Meta is still studying some of the achievements of DeepSeek, and

 the team hopes to eventually apply some of these advanced technologies to its own AI projects. Zuckerberg emphasized that Meta 

will continue to learn from technologies like DeepSeek and release its own open - source and free Llama AI model to promote the 

development of the industry. He also mentioned: "In the long run, significant investment in capital expenditure and infrastructure will 

still be a strategic advantage. Perhaps we will find something different in the future, but it is still too early to draw conclusions. For now, 

I firmly believe that the ability to build this kind of infrastructure will be a huge advantage."