The trendiest tech is a tiny silicon chip similar to video game graphics chips. It’s a chip for constructing AI systems like ChatGPT faster and cheaper.
These processors are now at the core of an AI revolution that might change the technology industry and the globe.
Nvidia, the premier AI chip provider, saw its shares rise over 25% last Thursday after forecasting a big revenue increase that experts claimed suggested surging sales. The corporation temporarily exceeded USD 1 trillion on Tuesday.
What are AI chips?
That’s hard to answer. “AI chips aren’t defined,” said Centre for Security and Emerging Technology research expert Hannah Dohmen.
In general, the word refers to computational gear that’s specialized to handle AI workloads, such as teaching AI systems to solve tough problems that might choke normal computers.
Game Origins
Three entrepreneurs started Nvidia in 1993 to advance computational graphics.
Within a few years, the business produced a GPU chip that performed numerous sophisticated graphics computations at simultaneously, speeding up video game production and play.
Parallel processing would help build games and AI.
Two University of Toronto graduate students developed a GPU-based neural network to win ImageNet 2012 by detecting picture images with considerably reduced error rates.
The success sparked interest in AI-related parallel processing, giving Nvidia and its competitors a new economic opportunity and giving academics significant tools to further AI development.
AI Chips
Eleven years later, Nvidia dominates AI chip manufacturing. The H100 GPU has 80 billion transistors, 13 million more than Apple’s most high-end MacBook Pro chip.
The H100 costs USD 30,000 online, as expected.
Nvidia doesn’t make these complicated GPU processors, which would need massive factory expenditures. Taiwan Semiconductor Manufacturing Co. and Samsung Electronics provide its chips.
Cloud-computing businesses like Amazon and Microsoft are major AI chip buyers.
Smaller organizations and organizations that couldn’t afford to construct their own AI systems can leverage cloud-based solutions for medication research and customer management by renting their AI processing capacity.
Others and competition
Parallel processing has numerous non-AI applications. For example, cryptocurrency miners bought most Nvidia graphics cards a few years ago to solve difficult mathematical problems for bitcoin payouts. Early 2022’s bitcoin market crash solved that issue.
Analysts predict stiffer competition for Nvidia. Advanced Micro Devices, which competes with Nvidia in computer graphics processors, might be a competitor. AMD has improved its AI processors.
Santa Clara hosts Nvidia. The company’s president and CEO is co-founder Jensen Huang.
