A fledgling startup aims to become one of Europe's first “AI computing” hyperscalers, with renewable energy playing a key role in its pitch to potential customers.
The AI gold rush has created an unprecedented demand for “computing.” “Computing” refers to the processing power, infrastructure, and resources required for tasks such as running algorithms, running machine learning models, and processing data. One company that has benefited greatly from this demand is Nvidia, which has emerged as a $3 trillion company driven by demand for GPUs (graphics processing units) and related AI hardware.
In parallel, an industry of cloud infrastructure providers has sprung up thanks to Nvidia, raking in buckets of cash in the process. In the US, we have seen the likes of Lambda and CoreWeave reach lofty multi-billion dollar valuations to expand their data center operations. Now, Finnish startup DataCrunch is throwing its hat into the ring, touting itself as one of the “few serious players” in the space with all its operations in Europe.
Finnish DataCrunch team. Image credit: DataCrunch
“GPU as a Service”
Founded in 2020 by CEO Ruben Bryon, DataCrunch, like its peers, sells GPUs “as a service” and promises to reduce the cost of AI processing. The company announced today that it has raised $13 million in seed funding, comprised of $7.6 million in equity financing from backers including ByFounders, J12 Ventures, and Aiven co-founder Oskari Saarenmaa. The remaining $5.4 million debt segment comes from Local Tapiola and Nordea.
While it's somewhat unusual for a seed-stage startup to raise such a large amount of debt, DataCrunch's move comes at a time when other companies in the space, such as CoreWeave, have also raised significant amounts of debt. For exactly the same reason. It's about using a physical asset (like an Nvidia GPU) as collateral to secure a loan, rather than transferring more equity.
It's also more efficient to secure a large bucket of capital this way, as the bank can simply take away the GPU if things get worst for DataCrunch. For those in control of the purse strings, it's much less risky than investing in a pure SaaS startup, for example.
“Given the business we are in, the main expenditure for expansion is capital expenditure. [capital expenditure] Bryon told TechCrunch. “This is a logical move and will allow us to raise additional funding as we grow.”
This new round brings the total amount of funding DataCrunch has raised since its inception to $18 million, and will go some way toward building out the infrastructure to support Nvidia's latest servers and clusters, including its shiny new H200 GPUs. This will lead to an expanded customer base that includes not only corporate customers such as Sony, but also individual AI researchers working at companies such as OpenAI.
“This has always been an important market for us, and I think this 'individual' market has been left behind by a lot of people,” Bryon said. “For me personally, it's important. I often use our unique service on the weekends, and have done so since the beginning.”
In fact, flexible, on-demand pricing makes it a much more attractive proposition for independent researchers and developers who need a bit of computing for personal or university projects.
“People who are studying for master's degrees or Ph.D.s. That's the demographic we want to stay connected to, because they're often the ones who are a few years away from achieving something really great. '' Bryon said.
Engage them now and reap the rewards later when you achieve great success. That's the general gist.
But there's no escaping the huge elephant in the room. This elephant in the room is one that all cloud companies must consider, and it will take an enormous amount of energy to power this AI revolution.
green machine
One of DataCrunch's “advantages” is the fact that its data centers are located in Helsinki, the capital of Finland, and in Iceland, where it has already been running on 100% renewable energy for years.
“In Helsinki, you can contract green energy from the grid,” Brion said. “And now one of our two data centers in Finland is recovering waste heat to heat Helsinki itself. In Iceland, where ambient temperatures are consistently low, the energy mix of the grid is already 100% % green. Therefore, Iceland is one of the most environmentally friendly places in the world for this kind of business to take place.”
This will be a big focus for the company going forward. The company will serve businesses all over the world, but will be primarily based in Northern Europe and Iceland. “Perhaps in the future, if we can find a suitable location that offers similar benefits in terms of the carbon footprint of our operations, we will consider Canada,” Brion said.
These “green” credentials are also expected to differentiate DataCrunch from other European rivals. Companies like France's FlexAI recently came out of stealth with $30 million in seed funding. And Nebius has recently risen from the ashes of Russian internet giant Yandex to become a public company again.
However, there is a trade-off here. Low latency is often one of the big selling points for AI computing providers, but DataCrunch doesn't necessarily fall into that bucket. This means that they are better suited for certain types of applications. workload.
“Our strategy is that just because we have 100 locations around the world doesn't mean we're going to be the absolute lowest latency provider,” Bryon said. “We're focused on computing that doesn't have very strict latency requirements, but still has good latency. Maybe not 10ms, but still around 100ms. It will be.”
It's also worth noting that DataCrunch's data centers are currently in shared “colocation” facilities, although the company has said it plans to begin construction of its own data centers in 2025. This will require significantly more capital.
“I want to get this company on track to go public, and we need access to more capital to continue expanding the company,” Brion said.