- The hyperscale capacity crunch is serving up enterprise customers for Lambda and other neoclouds
- Winning that business is one thing, but keeping it will take more than GPUs
- With a neocloud shakeout looming, Lambda is building for the long haul
Hyperscalers are leaving money on the table, and neoclouds are more than happy to pocket the change. Whether it’ll be enough to keep the lights on long-term is another story, but for now, the mismatch between supply and demand means good eating for companies like Nvidia-backed Lambda.
Amazon CEO Andy Jassy told a16z that AWS capacity is sold out for 2026, 2027 and a “good chunk” of 2028. While Amazon is spending $220 billion on CapEx in 2026, supply constraints are limiting how much new capacity it can bring online.
“I wish I could tell you that I thought the supply constraints would be just a 2027 issue. I think we’ll see them in different components for several years,” Jassy said.
It is against this backdrop that Lambda is looking to execute. Lambda’s Head of Cloud Services Dave Ward – who previously spent 18 years at AWS, including a year personally shadowing Jassy as his technical assistant – told Fierce the company is hustling to bring new data center online over the next 12-18 months. The goal, he said, is to scoop up the business that hyperscalers just don’t have the compute to handle.
But Ward acknowledged the trick isn’t securing these enterprise customers; it’s keeping them for the long haul.
The big question
“Right now, large hyperscalers and others are turning away large enterprises. And so it’s an opportunity to win that business,” Lambda’s Head of Cloud Services Dave Ward told Fierce. “But then the question is why would they stay with a neocloud in the long term? Why wouldn’t they just go back to a hyperscaler?”
The answer, Ward said, has three facets. First, Ward acknowledged that neoclouds will all need a baseline software platform to remain competitive. Second, he said they will need to differentiate themselves with specialized capabilities that are different from the common denominator-type services hyperscalers focus on. Think offerings like specialized storage options or certain stack optimizations. And third is an unbeatable operational performance muscle.
“When I was over at AWS, I remember losing customers potentially on price at some points and they would come back based on operational performance,” he said. “That muscle, that focus, the way you actually engineer differently for it is the difference between when a customer will stick with you or not.”
For Lambda, that operational push includes standardizing how it brings servers online and runs them across data centers. Ward said the company needs repeatable processes as it works toward scaling from megawatts to gigawatts. He also pointed to building more of its own data centers, alongside using colocation, as a way to differentiate its cost structure over time.
Getting its foot in the door early and working shoulder-to-shoulder with companies as they’re building their AI systems creates this “great flywheel” for both parties, he added.
Murky future for neoclouds
Lambda isn’t the only one facing questions about its future strategy. Fellow players including CoreWeave, Crusoe, Nebius, Vultr and others have also come under scrutiny as investors probe whether rising capital expenses will actually deliver the promised returns.
Not everyone is confident they will.
Gartner recently forecast more than 80% of neoclouds could be forced to abandon independent operations if AI revenues fail to offset rising CapEx costs. Elon Musk, who is looking to put a massive constellation of data centers in space, similarly predicted “death and consolidation” for a majority of neocloud players (for whatever you think that's worth).
Read more about neoclouds on FierceNetwork:
Wall Street’s anxiety catches up to neocloud CoreWeave
Agentic AI is driving CPU demand nearly 1:1 with GPUs, Vultr says
Gartner warns AI data center capex is ‘unsustainable’ as $3.5T write-down risk looms