A quirky group of startups that host and rent out the chips for AI tools undertook a fundraising effort last year.
From Iain MartinForbes contributor
W hen Vultr founder David Aninowsky began building his own data centers back in 2014. Customers only cared about cloud computing and the processors that supported it. Graphics processing units (GPUs) were the stuff of gamer forums, not corporate boardrooms. Now these chips are essential for developing artificial intelligence models, and leasing them has catapulted Vultr to a value of $3.5 billion.
The Florida-based startup raised $333 million from chip giant AMD and hedge fund LuminArx Capital earlier this month. Vultr has built more than 30 data centers around the world over the last decade and offers the advanced level of hosting typically sold by AWS, Google Cloud or Microsoft Azure. Renting time on GPUs has become the biggest growth driver. “AI is the fastest growing part of the infrastructure market,” said JJ Kardwell, CEO of Vultr Forbes.
It’s not just Vultr. According to a report, investors poured around $20 billion into 25 companies that lease access to GPUs last year Forbes Analysis of company documents and pitchbook data. That includes $8 billion in equity and over $12 billion in debt in the form of loans from Wall Street giants like BlackRock, Carlyle and Pimco.
It’s a fast-growing sector, but no one really knows what to call it yet: “Cloud GPUs”? “GPU factories”? Dylan Patel, an analyst at SemiAnalysis, chose something different. “New giants will emerge from the neoclouds,” he said Forbes.
The biggest winner so far has been Coreweave. The New Jersey-based startup had built an armada of GPUs for mining cryptocurrencies, but after the market crash in 2018, it switched to renting them out to startups working on AI projects. Last year alone, the company raised $1.75 billion in equity and $8.1 billion in debt, reaching a valuation of $23 billion.
The group of startups trying to catch up with Coreweave is a strange bunch. These include crypto refugees like Crusoe Energy, which began mining Bitcoin using gas flared from oil rigs, and Germany-listed miner Northern Data, which received a US$1.1 billion lifeline from stablecoin giant Tether -dollars to reinvent itself as an AI computing powerhouse.
This space also includes legacy data center companies such as Vultr and France’s OVH, which have transitioned from cloud computing to AI computing. Then there are players like Nebius, which emerged from the ashes of Yandex, the “Google of Russia,” with a Finnish data center, $2 billion in cash and a suspended Nasdaq listing. Shares resumed trading in October and earlier this month the company raised $700 million in a private placement from Nvidia and venture fund Accel to advance its new business – GPU rental. Smaller vendors include Runpod, which assembles smaller clusters of Nvidia chips, and Fluidstack, which finds buyers for data centers with unused chips.
Investors might be willing to overlook some of the strange backstories of this new group of unicorns, as GPU rentals largely generate hourly profits. “Previously, the payback period for purchasing a GPU was six months. It’s been a few years now,” Patel said.
“There’s a misconception that because Amazon is the Walmart of e-commerce, they have to be the Walmart of e-commerce, and they have to be the Walmart of cloud computing too. You really are the Niemann Marcus”
The boom in neoclouds is due not only to the shortage of AI chips, but also to massive price undercutting by giants like AWS and Oracle. Coreweave sells access to AI chip workhorses like the Nvidia A100 for $2.21 an hour. According to a pricing analysis by Neocloud competitor Paperspace, AWS charges $5.12 when renting GPUs by the hour. However, the Amazon-owned cloud giant offers significant discounts for long-term bookings. “There is a misconception that because Amazon is the Walmart of e-commerce, it must also be the same for cloud computing,” Vultr’s Kardwell said. “You really are the Niemann Marcus.”
Neoclouds are able to keep prices low because they sell “bare metal” GPUs without the bundled software and services typically offered by giants like AWS. While companies value the concierge experience, AI startups are often just looking for the lowest price. “AI factories are not investing in the software like AWS, Google and Microsoft are doing to make these clouds usable for whatever kind of computing power you need,” said chip analyst Karl Freund.
These pared-down, smaller challengers also have an advantage when it comes to getting their hands on AI’s most important asset: the GPUs themselves. Nvidia has backed Coreweave and Applied Digital, while chip rival AMD has backed Vultr. This chip shortage has helped tech giants like Microsoft and Oracle become some of Neocloud’s top customers. Microsoft warned in its latest earnings release that “capacity constraints” in its data centers were leading to slower growth in its AI business. And that’s how Coreweave came about.
Ironically, Neoclouds’ selling point for the so-called hyperscalers building airport-sized data centers may be their scalability. Coreweave has doubled its data centers from 14 to 28 in the last year; It has a $10 billion deal with Microsoft over the next five years. Chase Lochmiller, CEO of Crusoe Energy, said the company launched a renewable energy-powered data center in less than a year, while a hyperscaler would take at least three years. And Oracle has reportedly signed a $3.4 billion deal to lease a Crusoe AI data package…
