Master | moment | Getty Images
Chip stocks rallied last year as investors bet on the semiconductor sector’s central role in the global expansion of AI infrastructure.
But the renewed volatility around chip stocks has sparked debate about whether it is a sign of broader concerns about AI demand.
In interviews with CNBC this week, several AI executives scolded the idea that demand is slowing, although they acknowledged that companies are becoming more cautious about the costs of deploying AI.
“I think the demand for AI is almost unlimited,” says Pat Gelsinger Intel Playground Global’s CEO and now general partner told CNBC on Wednesday, adding that energy availability is “the only real limiter.”
“Because how much economic benefit does increased intelligence bring you? Almost infinite in every industry imaginable,” Gelsinger added.
Table of Contents
Data center and chip players report delivery bottlenecks
A number of factors have fueled market volatility around stocks related to chips and AI data centers. An announcement by Meta that it would sell its excess AI computing capacity partially contributed to the selloff. While Meta stock has been surging on the news, it raised questions about whether this was a sign that there is a broader overcapacity of computing power out there. Elon Musk’s xAI also rented out its excess capacity this year.
And this week, Samsung, one of the world’s largest memory chip makers, forecast a huge rise in profits, but its stock fell. After shares soared more than 360% in the last 12 months, the market was wondering how much further it could go.
None of these moves appear to have dampened demand for computing power and the infrastructure behind it.
“What we are experiencing in terms of demand is extraordinary. There is much more demand than we can meet, and that has been our experience for some time,” Marc Boroditsky, Chief Revenue Officer at I won’tsaid CNBC on Thursday. Nebius helps build data centers NvidiaGPUs.
Andrew Feldman, CEO of Brain systemssaid the example of Meta and xAI selling its excess capacity is a “unique” case.
“For the industry as a whole, demand for computing power far exceeds available capacity, and we’re lacking data centers. I think as an industry, we’re lacking a lot of inputs for computing power,” Feldman told CNBC on Wednesday.
Cerebras, which went public earlier this year, is one of many semiconductor startups trying to become major players in the data center market and challenge Nvidia.
Rebellions, another South Korean chip startup backed by Samsung and SK Hynix, reported similarly strong demand.
“Dynamics of AI infrastructure [is] still huge,” Rebellions CEO Sungyun Park told CNBC on Wednesday.
“I personally don’t think it’s the signal that says…all hyperscalers.” [are overinvesting] in infrastructure,” Park added in reference to the Meta and xAI news.
Lumentumwhich sells photonics and optical products for data center connectivity, said its products will sell out in the next five years.
“We’re trying to expand our capacity as much as possible to meet demand that we see five years from now at this point,” Lumentum CEO Michael Hurlston told CNBC on Wednesday.
Lumentum shares have risen about 600% in the past 12 months as investors pile into companies that are solving key bottlenecks in building AI data centers.
Corporate spending needs to be “rationalized.”
Another big debate surrounding AI trading is how much companies are willing to pay for the technology.
There was a period of so-called “tokenmaxxing” in companies, where companies would encourage their employees to use as much AI as possible, regardless of the outcome. The commonly used tools came from pioneering labs like OpenAI and Anthropic.
But companies are now more focused on AI’s return on investment, especially as these frontier models remain expensive compared to open source offerings from companies like DeepSeek or… Alibaba.
Nebius’ Boroditsky said that tokenmaxxing is only worthwhile if an organization gets a return on investment from it.
“The CFO who is lowering the hammer and slowing spending should actually be looking for value or valuemaxxing,” Boroditsky said, adding that AI should be used to create value that justifies spending.
“We’re now seeing a shift toward more rationalization. We’ve seen this with every technology cycle, and this rationalization will definitely sustain demand,” Nebius’ Boroditsky said.
While Frontier AI models are considered the most advanced, there are a variety of open source models that are similar in terms of performance and some that are less advanced. Different models have different capabilities that can be used for specific tasks.
Cerebras’ Feldman said that in the future, specific models will be used in specific situations. For example, boundary models can be used for more complex problems while offloading some workloads to others.
“I think it’s probably the case that you don’t need a huge bus to go to the grocery store,” Feldman said.
“Certain workloads migrate to one type of computing power and simpler workloads migrate to another, and I think as we learn and become more sophisticated in deploying AI, the same thing will happen.”
Choose CNBC as your preferred source on Google and never miss a moment from the most trusted name in business news.
https://www.cnbc.com/2026/07/12/ai-demand-chips-data-centers-stock-volatility.html
