A fast-moving AI startup, ready to scale.
Altrove is a Paris-based AI startup reinventing materials discovery — helping manufacturers find sustainable alternatives to rare earth elements (REEs) like cobalt and nickel. These materials are critical for EVs, smartphones, and defense systems, but global supply is volatile, expensive, and politically risky.
Altrove’s platform speeds up materials discovery using a combination of machine learning, simulation, and lab automation. That approach makes the work 100x faster than traditional trial-and-error methods — but it also demands high-performance, reliable compute.
Founded by a Cambridge-trained physicist and a second-time founder, and backed by Entrepreneur First (an accelerator backed by leaders from DeepMind, Hugging Face, Stripe, and more), the team raised €3.7M to bring their platform to market. But before they could scale, they needed serious GPU power.
The challenge: out of credits, and facing real demand.
Altrove initially ran on AWS, using ~$50K in startup credits to get their proof-of-concept off the ground. But once the core platform was working, they secured their initial funding — and with real customer delivery on the horizon, GPU infrastructure moved to the top of the list.
“As a startup in the tech space, AWS and other cloud platforms basically throw money at you,” said co-founder and CTO Dr. Joonatan Laulainen. “It’s great for early testing. But once the credits ran out, the costs were pretty brutal.”
They briefly considered buying their own hardware, but didn’t want to lock themselves into a setup they might outgrow.
“We didn’t know yet how our architecture or workloads would evolve,” said Laulainen. “It didn’t make sense to commit that kind of money upfront.”
The team started looking for a provider that could deliver high-performance GPUs — fast, flexible, and with transparent pricing they could trust.
The solution: dedicated H200s from Storj.
Altrove spoke with a few different providers, but quickly found their way to Storj as the better fit.
They knew what they were looking for: NVIDIA H200s, reserved for a year, with transparent pricing and reliable support. But finding them wasn’t easy. Many providers didn’t have the inventory — or wouldn’t commit to timelines. Storj stood out on all fronts.
“We looked around and said, ‘Who has H200s available now, and who’s actually willing to talk to us?’ Storj stood out right away.”
Hard-to-find H200s, reserved and ready.
Altrove had done the research — they knew H200s were the best fit for their simulation-heavy, high-memory workloads. But sourcing them on reserve was a challenge. Storj came through when others couldn’t, securing dedicated H200s and making it easy to move fast.
“What we do is considerably more performant on H200s… for high-memory tasks, that’s the better chip.”
Straightforward, transparent pricing from the start.
No long sales cycles. No vague promises. Just fast answers and clear terms — a refreshing contrast to vendors that dragged out the process or withheld details.
“It was fast and transparent… One of my bigger frustrations with others was that it was just really hard to get concrete information quickly.”
Responsive setup and sharp technical support.
Altrove’s setup required tight coordination and a few iterations, but the Storj team helped dial things in quickly — even across continents. Laulainen emphasized how helpful it was to work with a team that was both responsive and deeply knowledgeable.
“They were clearly very technically knowledgeable… that gave me confidence.”
Credibility that passed investor diligence.
Altrove brought their investors into the decision — and Storj cleared the bar. Knowing their infrastructure partner had staying power was critical.
“We involved our investors in checking out Storj… They’re not the kind of provider where you wonder if they’ll disappear next month. That really mattered to us.”
Results: compute power, flexibility, and speed — without the commitment trap.
With Storj, Altrove got what they needed: long-term access to high-performance GPUs, without getting locked into a rigid setup or overspending on post-credit pricing.
Key outcomes:
- Substantial savings compared to AWS and other clouds — not just on a per-GPU basis, but also by securing newer, better-matched hardware (H200s) without getting boxed into outdated configs
- No upfront capital investment — the team avoided a major hardware purchase at a moment when their architecture and workload needs were still evolving
- Room to adapt as they scaled — with reserved GPUs that could flex alongside real-world usage, Altrove didn’t have to overbuild or risk costly rework later
- Confidence that scaled with them — having a provider their investors could get behind helped clear diligence and accelerate go-to-market without second-guessing infrastructure
Facing a similar challenge?
If your startup is running into the same GPU roadblocks — credits drying up, unpredictable demand, or rigid pricing — we can help.
Book a 15-minute GPU setup review →
