Contents
Today, no AI data center can get past the chip manufacturer Nvidia. Nvidia is the third most valuable company in the world after Microsoft and Apple. There is no competition in sight. How was that possible?
In 1993, the film “Jurassic Park” thrilled audiences with computer-generated dinosaurs. At the same time, three young chip designers were thinking about how they could bring three-dimensional worlds for games to the PC. Special chips are required for the complex calculations.
In 1993, Chris Malachowsky, Curtis Priem and Jensen Huang founded the chip company Nvidia. Huang, 29, took on the role of CEO; He still runs the IT company today.
Permanent fear of competition
After initial difficulties, Nvidia brought the first graphics chip onto the market in 1997. Now you could move around in virtual space on the PC in games like “Star Wars”. The three of them achieved great success. Still, CEO Jensen Huang constantly worried that other chip makers might soon catch up. He was looking for new markets for Nvidia.
Graphics chips can perform many simple calculations in parallel. This makes them very fast for certain tasks. Accelerated computing – science could also benefit from this, Huang thought. In 2006, Nvidia opened up the graphics chip with the elaborately programmed CUDA software package for programming and thus for scientific calculations such as the processing of data from earthquakes or a CT scan – with a chip that was actually developed for games.
Great success thanks to foresight
Another area of application soon emerged: the accelerated calculation of neural networks that are behind AI applications such as ChatGPT. Huang contacted AI researchers around the world. Their feedback encouraged him to invest heavily in AI infrastructure, more than a decade before ChatGPT wowed the public.
Anyone doing AI research today can’t avoid Nvidia.
This foresight paid off: anyone doing AI research today can’t avoid Nvidia, says Gion Sialm, responsible for AI at the Graubünden University of Applied Sciences. There is no competition in sight in the foreseeable future. “There are hardware companies that may have better ideas, but hardware alone is of no use,” says the scientist.
Microsoft, Meta and Google depend on Nvidia
Accelerated computing requires a complex software infrastructure. Estimates suggest that Nvidia invested ten billion dollars in CUDA by 2017 alone.
Today, Nvidia supports completely different areas in addition to software for AI applications, from the simulation of entire factories to the development of AI-controlled robots to the simulation of quantum computers.
Jensen Huang’s risky bets on the future have paid off for Nvidia. For customers, however, success also has disadvantages: higher prices without competition. Even large IT companies such as Microsoft, Meta or Google now find themselves dependent on Nvidia from which they cannot escape so quickly.