Jensen Huang said he told Nscale to "Come, you guys, do a good job, work around the clock" and added, "That was it." The UK-based neocloud, backed by Nvidia, has run into holdups, breakdowns, and limited power capacity at certain locations, along with investors pushing back on its $30 billion IPO valuation target. The company did not provide comment right away when asked.
He was trying to bolster confidence in neoclouds, a slice of AI infrastructure that has cooled after a run of stumbles. These firms build AI-ready data centers, often financed in part by Nvidia and packed with its chips. Recent setbacks include Australian player Firmus Grid dropping a planned $30 billion IPO and South Korea's Elice having a difficult trading debut.
Huang said the group can get through the snags because AI computing demand remains strong. "I'm quite confident that they will find solutions, and their solution starts with the fact that they have excellent customers, demand is really high, and they just need working capital," he said at a press conference during Nvidia's GTC event in Berlin.
He also noted neoclouds are assembling some of the largest and most complex systems while dealing with construction, supply chain and delivery hurdles. Where data centers once took three years to build, "Now, if you don't get it up and running in six months, everybody's up in arms."
When technology and markets intersect it helps to get clear context, so sign up for the free Market Briefs
Nvidia's chips have fueled the AI boom, and Huang has been in Europe for more than a week, signing agreements and telling political and business leaders that AI can be safe and sustainable. His upbeat view meets growing concern about the massive spending required for AI infrastructure as borrowing costs climb. AI stocks have slid over the past two weeks, and the Philadelphia Stock Exchange Semiconductor Index just posted its longest losing streak since 2008.
At GTC Berlin, Nvidia unveiled fresh data center gear, software, models, and partnerships. A headline product was DSX Pod, a containerized unit that functions as a mini data center for companies that want to get hardware running quickly. Nvidia said each container arrives with server hardware, cabling and plumbing preinstalled and tested, and then simply requires connection to power and integration with other units.
Huang's case rests on demand, customer pipelines and bridging today's build frictions with working capital. The watch items now: whether neoclouds can bring facilities online in about six months as many now expect, secure sufficient power, and untangle supply hiccups while funding keeps pace with ambition.
Staying informed about complex infrastructure and finance makes decisions easier, so join the free Market Briefs newsletter
