ChatGPT Is Driving Strong Demand For Advanced Processors Among Chip Makers

Artificial intelligence systems that generate text with little to no prompting, demand colossal computer power to run, and promise a lucrative new revenue stream have chip makers buzzing.

‍

Analysts predict that the new tools might generate tens of billions of dollars in net annual sales for semiconductor producers if they are broadly adopted.

‍

Since the introduction of ChatGPT, a chatbot developed by San Francisco-based OpenAI, late last year, interest in so-called generative AI has exploded. The technology has captured consumers' attention by providing convincingly accurate, if occasionally unreliable, responses, which has helped it secure billions of dollars from Microsoft Corp. Together with other investors, Microsoft -0.46%

‍

The top executive of America's largest chipmaker by market value, Nvidia Corp., Jensen Huang, claimed that technology has reached a turning point. During the company's quarterly earnings presentation on Wednesday, he said, "Generative AI's versatility and capability has triggered a sense of urgency at enterprises around the world to develop and deploy AI strategies." He also unveiled a new cloud-computing initiative to take advantage of the business opportunity.

‍

According to him, organizations are quickly resetting their business expectations due to the interest in these AI solutions. There is little doubt that the past 60 to 90 days have significantly altered any expectations we had for this year at the beginning of the year.

‍

The thrill comes at a time when the semiconductor sector is struggling to recover from a severe slowdown as sales of personal computers, smartphones, and other electronics sag. As a result of consumer and business spending cuts brought on by recession worries, the majority of chip manufacturers have reported sluggish sales.

‍

In the unglamorous world of data centers, where tools like ChatGPT do calculations and output findings, Nvidia is without a doubt the market leader in chips used for AI. An Omdia estimate states that as of 2020, it accounted for 80% of these AI processors.

‍

Nevertheless, other chip manufacturers want in on the action because there is so much money at stake.

‍

Apple Inc. According to CEO Pat Gelsinger, his company has a wide range of chips to address the generative AI opportunity, including specialized chips designed for AI computation, graphics chips for data centers, and a new generation of data center central process units, or CPUs, which he claimed were effective in AI work.

‍

According to him, as AI is more incorporated into all applications, "the performance we anticipate will become much more of the mainstream of computing."

‍

The maker of AI-specific CPUs, graphics chips, and other gear, Advanced Micro Devices Inc. (AMD -0.21%), is also banking that big cloud computing firms, which power many of the computations vital to the technology, would make significant semiconductor investments. According to AMD CEO Lisa Su, that business should start to grow more significant starting in the next year.

‍

By 2027, generative AI may boost the market for AI chips by $20 billion annually, predicts Bank of America analyst Vivek Arya. According to him, Nvidia should be able to keep at least 65% of the market for AI chips.

‍

This month, Google, a division of Alphabet Inc., gave a sneak preview of Bard, a domestic ChatGPT competitor. China's Baidu Inc. is creating Ernie Bot, an AI-powered chatbot that will debut the following month and be comparable to ChatGPT. Microsoft already offers a small taste of ChatGPT to users through the Bing search engine results.

‍

At least in the short future, Nvidia may be best positioned to profit due to its leadership in AI. The business took the lead by enabling software developers to take advantage of AI-friendly features of its graphics chips starting approximately 15 years ago. According to UBS analysts, the company's chips are now the only items that can be used to build large-scale AI language systems. They also estimate that ChatGPT needs about 10,000 of the company's graphics processors to train.

‍

After making a projection of $1 trillion or so around a year ago for its business ranging from delivering chips for videogames to autos, Mr. Huang hinted the company may alter its outlook for the size of its potential market next month. 

‍

We "are probably going to arrive at that [market size] sooner rather than later," he said, "because of the extraordinary capabilities and versatility of generative AI and all of the convergence discoveries that came toward the middle and end of last year." Without a doubt, this is a significant moment for the computer industry.

‍

By beginning to provide a cloud computing service for businesses to create generative AI chatbots and other tools using its hardware and software, Nvidia is attempting to get there faster. The service, which would be provided by reputable cloud computing firms, intends to remove entry hurdles for the expansion of AI application in the commercial world.

‍

Nvidia stated that it is collaborating on generative AI tools with consumer internet companies, startups, and all of the main cloud computing providers, including Amazon.com Inc., Microsoft, and Google.