Expanse Raises $5.3M to End AI Compute Guesswork

Jan de Vries · 2026-09-16

Ever launched an AI job only to watch it crash because you guessed the compute wrong? Or worse, over-allocated and burned through your GPU budget? You're not alone. That's exactly the problem Expanse is tackling, and they just raised $5.3 million to do it. ### The Seed Round That's Turning Heads Expanse, an AI infrastructure company, announced a €4.5 million ($5.3 million) Seed round led by Crane Venture Partners. PXN Ventures, AIP Seed, and angel investors—including former DeepMind researchers—also joined in. The money will go toward expanding the engineering team, speeding up product development, and getting the tech into more hands. Why the excitement? Because Expanse is going after a pain point that every AI team knows too well: the educated guess. "Today, every AI workload begins with an educated guess," says Ismaeel Bashir, co-founder and CEO. "Engineers shouldn't have to predict exactly how much compute their code will need before they press run. The machine should carry the uncertainty, not the person." That's a refreshing take. Instead of forcing humans to become fortune tellers, Expanse lets the software do the heavy lifting. ### Why This Matters Right Now Timing is everything. Expanse's round lands amid a wave of 2026 investments aimed at making AI compute more accessible and efficient. Here's a quick snapshot of recent deals: - London-based Callosum raised €85.4 million to route workloads across different AI models and chips. - Vienna's Ora Computing secured €3.5 million for model compression and inference efficiency. - Helsinki-based Verda raised €100 million for AI cloud infrastructure. - London-based Nscale closed a €1.7 billion Series C for vertically integrated compute infrastructure. - Other rounds: €16 million for Switzerland's AI Infrastructure Capital AG, €5 million for Danish startup Velatir, and €1.7 million for Prague cloud platform Zerops. Together, these 2026 financings total roughly €1.91 billion—or about €1.92 billion including Expanse—though Nscale's mega-round skews the number. The UK is clearly a hotbed, with Nscale and Callosum leading the charge. Callosum, like Expanse, is building software to allocate AI workloads more efficiently. And just a day before Expanse's news, Crane Venture Partners announced €419 million across four new investment vehicles. That's a lot of dry powder looking for smart infrastructure bets. ### How Expanse Works Its Magic Founded in 2025 by four University of Edinburgh graduate engineers, Expanse isn't your typical monitoring tool. Traditional observability platforms tell you what happened after a workload finishes. Expanse works before execution even begins. Installed directly into your cloud or on-premises infrastructure, its models analyze workloads without code or telemetry ever leaving your environment. That means you can optimize performance while keeping data sovereignty intact. The software predicts what an AI workload needs before it runs, helping you reduce over-allocation, prevent failed jobs, and squeeze more value from the GPUs you already own. > "The future of AI won't be defined only by who builds the biggest clusters, but by who uses them most intelligently," says Scott Sage, co-founder and Partner at Crane Venture Partners. "Expanse gives organisations something they don't have today: certainty. By predicting exactly what every workload needs before it runs, customers recover existing capacity, reduce waste and extend the life of some of the world's most valuable infrastructure. We believe that's a new and important category in AI infrastructure." The founding team's background is impressive. While at the Edinburgh Parallel Computing Centre (EPCC), Bashir developed a multimodal HPC resource prediction system. He and co-founders Nikodem Bieniek, Eren Mendi, and Yafet Melake realized that even the most sophisticated compute environments still relied on manual estimation. They knew there had to be a better way. ### What This Means for Your AI Strategy If you're running AI workloads at scale, you've probably felt the sting of wasted compute. Expanse's approach could be a game-changer. By predicting resource needs upfront, you can avoid the costly cycle of trial and error. That means faster experimentation, lower bills, and more reliable results. Plus, with data sovereignty a growing concern, Expanse's on-premises-friendly design is a smart move. You get the benefits of AI-driven optimization without shipping your sensitive data to a third party. The broader trend is clear: investors are betting big on infrastructure that makes AI more efficient. Expanse is now part of that wave. Whether they become the go-to solution for compute prediction remains to be seen, but they're certainly worth watching. For now, the company plans to use the fresh capital to grow its team and accelerate product development. If you're tired of guessing how much compute your AI needs, Expanse might just be your new best friend.