Unlock Editor’s Digest for free
FT editor Roula Khalaf picks her favorite stories in this week’s newsletter.
Nvidia CEO Jensen Huang has long argued that the world’s data centers will need to be completely redesigned to handle the demands of generative artificial intelligence. He claims it will take $1 trillion over the next four to five years to train and operate new AI models, essentially doubling the amount already invested in digital infrastructure.
Nvidia itself benefited most visibly. Its meteoric rise in the stock market has made it the world’s most valuable company.
However, the flurry of AI-related earnings announcements and deals over the past two weeks has also provided encouraging evidence that the boom sparked by the launch of OpenAI’s generative AI chatbot ChatGPT is catching on. Of course, there’s no telling whether the spending wave will be sustainable or big enough to justify a huge rally in tech stocks, but it should at least bring some comfort to the bulls.
Shares of chip maker Broadcom, for example, have jumped more than 20 percent since it announced its latest revenue boost driven by artificial intelligence. Another such rise, and it would join a limited group of tech companies valued at more than $1 trillion, more than six times what they were worth half a decade ago.
Much of the growth comes from demand for AI accelerators, chips that Broadcom custom designs for customers like Google to speed up their AI calculations. But its explosive growth also points to the more important role that high-speed networks play in data centers.
The sheer amount of information needed to train and run AI models has required much faster connections between individual processors and between different machines running in data centers. Broadcom CEO Hock Tan estimated that networking will account for 40 percent of his company’s AI chip sales by the end of this year.
Meanwhile, shares of cloud latecomer software maker Oracle jumped 17 percent on news of a deal to train OpenAI’s large language models on its cloud infrastructure. The deal involves bringing OpenAI partner Microsoft’s Azure cloud service to Oracle’s giant new data center — a relationship between the tech industry’s two arch-enemies that would have been unthinkable once.
Elsewhere, even Hewlett Packard Enterprise, which seemed to have missed out on the growing demand for AI servers that has lifted rivals Supermicro and Dell, has finally hit Wall Street. Its shares rose 24 percent after earnings as investors reassessed its position in the AI ​​boom.
As news like this has fueled hopes that the AI ​​boom is spreading to more vendors, a few things are becoming clear. One is that the impact appears to be widespread, covering many different parts of the “technology stack”—the hierarchy of components from chips to software that are needed to run today’s complex IT systems.
Nvidia is still in position to be the biggest winner by far. Most of its sales come not from individual chips, but from entire servers, often networked together in complete racks. Squeezing out the best performance comes from tuning every element of these systems to work together, using Nvidia’s own proprietary technologies in areas like networking.
Nvidia’s biggest customers are desperate to reduce their dependence on the company, pushing for new standards in everything from networking to AI software that would allow other competitors to emerge. But these initiatives will take time.
The biggest tech companies are also expanding their direct involvement in more parts of the infrastructure required by AI. A key part of Apple’s AI announcement last week, for example, included reports that it was designing its own servers, reportedly based on its own chip designs. Apple has already taken control of most key components in its phones: a similar move is likely in the data center as AI demands force it to offload more of its customer data processing back to its own devices.
One consequence of such moves is that vendors’ business models have had to adapt, with leaders such as Broadcom playing a supporting role as customers take more control. The so-called “hyperscalers” – the largest cloud companies – also bring a greater share of the total demand, which leads to dependence on a narrower base of large customers. This will increase the vulnerability of suppliers in any downturn. But for now, Wall Street is fixated on how many tech ships will be lifted by the rising tide of generative artificial intelligence.
richard.waters@ft.com