Addressing one of the most persistent critiques of the current artificial intelligence boom, CoreWeave CEO Michael Intrator pushed back against the narrative of a “circular AI economy” in an appearance at the Fortune Brainstorm AI conference in San Francisco.
While skeptics often point to the tangled web of investments between chipmakers, cloud providers, and AI startups as a financial bubble, he argued that deep industry collaboration is the only viable response to a historic supply chain crisis.
Circular is “the incorrect way of looking at it,” Intrator told Fortune Editorial Director Andrew Nusca, reframing the dynamic not as financial engineering, but as logistical necessity. “It’s a lot of companies working to address an imbalance that is distorting the globe.”
The concept of the “circular economy” in AI suggests that revenue is merely being recycled between a handful of tech giants—such as Nvidia investing in CoreWeave, which in turn uses that capital to buy Nvidia chips. However, Intrator described the market conditions as a “violent change in supply demand,” adding that the only way to navigate such volatility is “by working together.”
The ‘physical bottleneck‘
According to Intrator, the primary constraint facing the AI sector is not funding or policy, but “a physical bottleneck associated with getting … the most performant compute into the hands of the most cutting-edge players.” This scarcity forces companies to cooperate in ways that may look insular to outsiders but are essential for survival, he insisted.
The CEO recounted a recent conversation with a mining company boss, whom he declined to name. Intrator said he learned just how deep the supply chain is being impacted: “two levels deeper,” down to the raw metals and copper required to build the infrastructure. Intrator noted that the executive specifically requested industry-wide cooperation to meet production needs.
The mining CEO explained that to get out of this jam, “we need to work together as a group.” If he said the same thing about the AI space, Intrator reasoned, “I get accused of being in a circular economy … So that’s all I’ll say on the circular economy is, like, you do that by working together.”
Critics warn that if a firm like CoreWeave cannot roll its debt or loses a key client, lenders could dump large volumes of used GPU chips into secondary markets, hitting hardware prices and rippling through the AI supply chain. But Intrator described a rapid, even violent escalation of demand.
Managing ‘relentless’ demand
CoreWeave, which specializes in parallelized computing solutions essential for AI, sits at the center of this storm.
“The demand from the most knowledgeable, most sophisticated, largest tech companies in the world is relentless,” Intrator said. “That’s what the trend that matters to me.”
This rapid expansion has come with volatility. Since its IPO, CoreWeave’s stock has seen significant fluctuation, a phenomenon that Intrator attributed to the market adjusting to a disruptive business model challenging the traditional cloud dominance of major tech players. Despite the “seesawing” stock price, Intrator noted that the company has been successful, with the stock trading around $90, compared to an IPO price of $40.
He also addressed concerns regarding customer concentration. While he acknowledged that CoreWeave was previously reliant on Microsoft for 85% of its revenue, he said aggressive diversification efforts mean that no single customer now represents more than 30% of the company’s backlog.
The super-cycle view
Intrator urged investors to look past short-term execution hiccups, such as a data center opening delayed by a week, which he said caused “bedlam” among myopic observers. Instead, he views the current landscape as a “macro super-cycle,” where the fundamental shift from sequential to parallelized computing is opening up computational power at an order of magnitude previously unimagined.
Ultimately, the collaboration that critics decry is the mechanic that is moving the industry forward, Intrator maintained. “The reasons that you have challenges in delivering that compute is because of policy… because of physical infrastructure … because of energy,” he said. “You do that by working together.”
