NVIDIA responds directly: What are the strengths of GPUs over ASICs?
NVIDIA believes that commercial GPUs have platform advantages, while custom ASICs are limited in application scale and scope because they are only suitable for specific applications. In the future, AI will become part of national infrastructure, and NVIDIA's general GPU solutions and software support will provide global support in a way that custom ASICs cannot achieve
During the Q&A session at the "Spring Festival Gala of the Tech World" CES exhibition, NVIDIA clearly pointed out the advantages of GPUs over ASICs.
According to UBS's latest summary report, NVIDIA stated that commercial GPUs have platform advantages, including complete hardware and software support, which makes GPUs more competitive globally; in contrast, custom ASICs are limited in application scale and scope because they are only suitable for specific applications.
In addition, NVIDIA also mentioned at the exhibition that in the future, AI will shine in multiple fields such as automotive and robotics, with "physical AI" being the next key area, and quantum computing will not become mainstream technology for at least the next fifteen years.
What are the strengths of GPUs over ASICs?
NVIDIA stated that machine learning is the direction for future development, which means that accelerated computing (improving computing efficiency through hardware acceleration) will replace general computing as the development direction in the field of computing.
Commercial GPUs, such as NVIDIA's GPUs, are general-purpose hardware designed for multiple tasks and applications, offering strong flexibility and scalability; while custom ASICs are hardware tailored for specific tasks, which, although highly efficient for specific tasks, lack flexibility and scalability.
Therefore, NVIDIA believes that commercial GPUs have platform advantages, including complete hardware and software support, which makes GPUs more competitive globally; in contrast, custom ASICs are limited in application scale and scope because they are only suitable for specific applications.
NVIDIA further stated that in the future, AI will become part of national infrastructure, and NVIDIA's general-purpose GPU solutions and software support will provide global support in a way that custom ASICs cannot achieve.
NVIDIA: Robotics technology will be the next key application
NVIDIA believes that in the future, AI will shine in multiple fields such as automotive and robotics, with "physical AI" being the next key area, and robotics technology will be the first major application in this area.
Regarding the automotive sector, NVIDIA stated:
In the future, every car company will need two factories—one for manufacturing cars and another for developing AI.
As for quantum computing, NVIDIA believes that while quantum computing excels at handling small data problems (such as random number generation and encryption), it performs poorly with large data problems. Although the company currently collaborates with most major quantum computing companies, NVIDIA believes that quantum computing will not become mainstream technology for at least the next fifteen years.
UBS also maintained NVIDIA's target price at $185 per share in the report, which represents a 32% upside compared to yesterday's closing price of $140 per share.
