At the launch of CUDA on GeForce GPUs, NVIDIA CEO Jensen Huang discussed a critical moment, a gamble that could have put his company at risk.
CUDA: A Risky Bet for NVIDIA CUDA, launched in 2006, has become a fundamental element for NVIDIA, according to Jensen. This strategic framework allowed the company to establish itself as a major player in the AI race. Initially, NVIDIA aimed to distance itself from the image of a “GPU company,” aspiring to become a complete provider of computer solutions.
“The more we became a computing company, the less we could specialize,” said Jensen.
The design of programmable shaders paved the way for a new era for NVIDIA. These shaders were intended to break free from the traditional constraints of CPUs to explore other areas of computing. Although the market had a limited reaction to the launch of CUDA, Jensen continued to believe in its potential to revolutionize the industry.
Painful Beginnings in Programmability To succeed, Jensen emphasized the importance of supporting computation FP32, a turning point that attracted researchers to GPUs for intensive workloads. However, at the time, NVIDIA knew that the return on investment would be long and risky.
“This decision was costly, but it was the price to pay for the ambition of being a computing company,” said Jensen.
The path to CUDA’s success took a decade, marked by the patience of Jensen and his development team. Today, CUDA is the engine driving NVIDIA’s success in many market segments.
“NVIDIA is truly the company that GeForce has built – it is thanks to GeForce that CUDA has spread widely,” Jensen noted.
Jensen’s determination established CUDA as a standard in the industry, forever changing the perception of GPUs.
[Fact Check: CUDA is a parallel computing platform and programming model developed by NVIDIA for general computing on GPUs.]
[Context: NVIDIA’s CUDA technology has been instrumental in the field of artificial intelligence and high-performance computing.]
