Intel bets on AI beyond data centers

U.S. semiconductor giant Intel aims to make AI capabilities commonplace in products from computers to edge computing to software, and some companies are already moving their data from public clouds to private environments, said Alexis Crowell, vice president and chief technology officer for Intel's Asia Pacific and Japan operations.

“Our strategy is to make AI ubiquitous,” Crowell said in an interview. “If there was only one solution in the world that could compute AI, we'd be stuck. If all chips and software could compute AI, then we have flexibility.”

Growing interest in AI services and the massive computing power they require has led to a surge in data centers around the world. In Asia, too, companies such as Alphabet, Microsoft, and Amazon are spending billions of dollars building cloud services and other computing infrastructure. Morgan Stanley predicts that Asia will account for more than one-third of the world's data center capacity by 2027, spurring $100 billion in investment over the decade.

But the pace of demand for traditional data centers is likely to slow as companies and organizations try to balance using public cloud infrastructure with keeping data private, Crowell said.

“I don't think we're going to see a big explosion and continue on a steep trajectory,” Crowell said. “Some of the data I'm seeing suggests that businesses are working on a recovery. [outsourcing of IT] “Either for data privacy reasons or cost control reasons.”

This is in line with a recent report from research firm IDC, which predicts that by next year, 75% of data generated by businesses worldwide will be created and processed at the edge, outside of traditional data centers or the cloud.

Intel is primarily known for its dominance in the market for central processing units, the brains of personal computers, and the servers that run corporate networks and the Internet.

The Silicon Valley company's other main business, supplying chips to data centers, has struggled in recent quarters. AI chips are eating into the budgets of data center operators, but Nvidia dominates the space and Intel has a smaller foothold. Intel's data center and AI division revenue rose about 5% to $3 billion in the first quarter, while overall revenue rose about 9% to $12.7 billion.

Intel is looking to change that as part of its efforts to become the world's second-largest contract chipmaker by 2030. Last month, the company unveiled the third generation of its AI chip, Gaudi 3, which the company says will surpass Nvidia's H100 in training speed and power efficiency, helping it hit $500 million in AI chip sales in the second half of 2024. The company said the chip will be available to companies such as Dell Technologies, Hewlett Packard Enterprise and Lenovo this quarter.

“There are use cases where public cloud infrastructure makes sense for businesses, their customers and their environments. [and] No, we [have] “For sovereignty reasons and other reasons, all of this data needs to be kept completely private,” Crowell said.

“It's just a matter of companies and organisations, including universities and governments, figuring out what the right balance is,” she said.

Write to Kimberley Kao at kimberley.kao@wsj.com.