The moment a major fintech company files for an IPO, the cracks in its financial modeling infrastructure become impossible to ignore. For three executives at Klarna, navigating the Swedish payments giant's September 2025 public listing, those cracks looked less like minor flaws and more like fundamental design problems—the kind you can't patch with another macro or pivot table.
That experience, as it turns out, became the founding thesis for Galdera Labs.
The Stockholm startup announced a €1.5 million pre-seed round on March 26, 2026, led by J12 Ventures with backing from Antler and a roster of strategic angels. It's a modest raise by today's inflated standards, but the company's ambition is anything but: Galdera wants to kill the spreadsheet.
Or rather, to replace it with something the founders believe can actually handle the complexity of modern finance—what they're calling "reasoning infrastructure" for the CFO's office.
The Excel Problem Nobody Wants to Admit
The three co-founders—who collectively bring stints at Deloitte, Bain & Company, Accenture, and BlackRock before landing at Klarna—spent their time there building FP&A infrastructure across 26 markets. They know the pain points intimately. Models get rebuilt from scratch. Context evaporates between versions. Assumptions live in someone's head or, worse, buried in a comment box that half the team never reads.
"Financial modeling has been stuck in a world of disposable spreadsheets," said Emmet King, Managing Partner at J12 Ventures, in a statement that probably resonates with every finance professional who's ever searched desperately for "final_final_v3_ACTUAL.xlsx."
The broader issue, according to Galdera's pitch, is that Excel was never designed for the way AI needs to interact with financial data. You can bolt on chatbots and automation, sure. But if the underlying structure treats models as static artifacts rather than living systems, the gains remain marginal.
Memory as Infrastructure
What Galdera is building—or says it's building, given the company is still in early customer rollout—sounds less like FP&A software and more like a reimagined database with a calculation engine grafted on top. The platform centers on what the team calls a "semantic memory layer," which links financial figures not just to formulas but to the business context that produced them: the assumptions, the decision history, the why-did-we-model-it-this-way.
Users can query the system in natural language. Run scenarios for market entry, pricing shifts, headcount expansion. Watch impact ripple through the model in real time. The platform integrates with existing data warehouses—Snowflake, BigQuery, the usual suspects—but treats the model itself as something closer to institutional memory than a monthly deliverable.
Whether this actually works at scale, of course, is the open question. The company hasn't named customers publicly, describing them only as "some of Europe's most forward-thinking finance teams," which could mean anything from stealth-mode unicorns to a handful of pilot partners willing to take a flyer on unproven technology.
Timing and Tailwinds
Tobias Bengtsdahl, a partner at Antler, framed the investment around a familiar pressure point: finance teams being asked to deliver strategic insights without adding headcount. "AI-native infrastructure that captures institutional knowledge is no longer optional," he said—a view that's gained traction as AI skills have crept into job descriptions across the function.
A study cited by CFO Dive found that nearly one in three finance roles posted between January 2025 and January 2026 now explicitly require AI capabilities. That's a striking shift in just twelve months, though it's worth noting that "AI skills" on a job posting can mean anything from prompt engineering to basic familiarity with Copilot.
Galdera is entering a market that's suddenly crowded with well-funded competitors. Datarails pulled in $70 million this past February to expand its AI agent offerings. Pigment raised a hefty $145 million Series D back in April 2024, and Abacum closed a $60 million Series B in June 2025. All of them are chasing some version of the same thesis: that AI can finally make financial planning less painful.
The difference, Galdera's founders argue, is architectural. Most tools are layering AI onto spreadsheet logic. Galdera wants to start from scratch—models designed for continuous reasoning, not monthly reconciliation.
The €1.5 Million Question
Whether European finance teams are actually ready to abandon Excel entirely is another matter. Spreadsheets have remarkable staying power, less because they're good at what they do and more because they're familiar, flexible, and embedded in decades of workflow muscle memory. Convincing a CFO to rip out that infrastructure and replace it with something untested requires more than a compelling pitch deck.
It requires proof. And for now, Galdera is still in the proving stage.
The funding will go toward platform development and expanding that early customer base, which means the next year or so will be critical. Either the semantic memory layer becomes the foundation for a new category of financial tooling, or it joins the long list of ambitious ideas that looked better in theory than in practice.
For three former Klarna executives who spent months preparing investor disclosures for an IPO, the stakes are probably familiar. The difference is that this time, they're building the infrastructure they wish they'd had.
