On theCUBE Pod: AI bubble debate heats up and neocloud earnings challenge doubters
SiliconANGLE Devony Hof ● Covered by 7 sources
AI bubble talk is getting louder, but one analyst says Nvidia’s $500 billion push could keep it afloat. Neocloud earnings and Cisco’s AI role are making the doubters look a bit too smug.
Based on reporting by SiliconANGLE, Devony Hof — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
TheCUBE Pod picked a timely fight this week: is AI already in a bubble, or is the buildout still early enough to outrun the crash? Dave Vellante says the bubble risks are real, but he thinks they may take longer to bite because Nvidia Chief Executive Jensen Huang has committed $500 billion to independent financing platforms for AI infrastructure. If that money keeps the buildout moving, the whole thing can stay inflated a little longer than the skeptics expect.
John Furrier pushed the argument in a different direction. He said AI demand is not just about hardware, but about what the hardware produces, especially inference and agent-driven workloads. In his view, robotics, high compute and falling prices could expand demand faster than supply catches up. That would not erase bubble risk. It would just make the market look more like a giant consumption phase than a simple boom and bust.
Vellante’s case is more cautious. He pointed to the usual bubble markers: prices rising quickly, capital piling in, and the danger that financing dries up before the end users show up to pay for it all. He also said the AI supply chain is tight in the places that matter most, including high-bandwidth memory, advanced packaging, network fabric, power and site readiness. Those bottlenecks could delay relief on the supply side and stretch out the whole cycle.
The deeper warning is that Huang’s financing push turns AI compute into collateral. If demand stumbles, the pressure does not stay neatly inside one company. It ripples through semiconductor suppliers, neoclouds and infrastructure investors. Vellante described it as two clocks running at different speeds: the fast IT clock for compute, storage and networking, and the much slower clock for data centers, power and regulatory approvals.
That slow clock is exactly why Furrier sees risk in the neoclouds, even as he says the user demand looks real. And the numbers from CoreWeave and Nebius help explain the optimism. CoreWeave reported $2.6 billion in quarterly revenue, while Nebius said revenue reached $582.3 million. Cisco also got pulled into the AI conversation after beating earnings expectations but seeing its stock fall after hours, which Vellante and Furrier called a mistake by investors who may be missing how central networking still is to AI.
My take — AI-written commentary, not fact-checked reporting
The market loves AI as long as it can count the racks and ignore the bills. That’s usually the cue to stop clapping and start reading the plumbing invoices. The real tell here is not whether AI is hype; it’s whether the financing machine has become big enough to postpone reality, which is a very Silicon Valley way to define “healthy.”
Read more about this at: SiliconANGLE