Endra Raises $50M to Automate Data Centre Design - Ventureburn

Stockholm-based AI startup Endra announced that they have raised a $50 million Series A round. Andreessen Horowitz (a16z) led the funding round to expand Endra’s automation platform for mechanical, electrical, and plumbing systems.

Notion Capital and Norrsken VC, both earlier investors of the startup, participated in the round, bringing the total funds raised to date for Endra to $75 million in its first two years of trade.

This huge cash infusion underscores an emerging pattern among the biggest of the big venture investors to finance the foundational, heavily taxed, and complicated backend engine that propels this AI hype cycle. Capital is shifting fast from consumer software and cloud apps to physical infrastructure problems. Manual engineering will determine when and how fast power-hungry data centres get built.

Overcoming the 25-Year Engineering Bottleneck

For the past 20+ years, the design process for laying out the sophisticated internal infrastructure of today’s buildings has been relatively unchanged. Although architectural design transitioned to 3D visual representations long ago, the logistics of placing large power transformers, cooling loops, electrical trays, and fibre backbones inside a building shell have, for the most part, been a labour-intensive task. Engineers often spend months averaging millions of coordinates, referencing complex national and local electrical codes.

This automated process has become a killer bottleneck for the global data centre industry. As hyperscale computing farms grow 3x over the next decade for AI needs, construction timelines are stretching to the breaking point. MEP engineering shortages have grown 3x, putting huge delays on deployment, while tech giants are begging for more processing capacity.

Endra intends to efficiently and systematically remove this friction by replacing weeks of legacy human layout changes with truly custom, optimised generative algorithms. Its dedicated design platform launched inside Autodesk’s industry-standard revit native application base used by worldwide engineering firms, enabling engineers to migrate their workflows without discarding existing files or tooling libraries.

More News: Invisix Raises €20M to Advance AI Chip Manufacturing

The Architecture of Spatial AI and Physics

To generate code-compliant architectures without the inaccuracies typically associated with large language models, Endra implements a layered hybrid system. Although it uses language models to read and translate building codes into rules, a proprietary 3D geometry engine performs layout generation using deterministic algorithms.

The resultant transition in performance is enormous. Upon importing an architect-scaled 3D model, the system seamlessly inserts conflict-free 3D schematics indicating lighting grids, multi-layer wiring systems, ventilation conduit, and fire system devices. In addition to this, the framework automatically records all relevant computational records, agent records, and asset tables. Company data shows the Endra platform could solve and schedule a ten-week UK engineering project in just eight hours.

Targeting the Multi-Billion Dollar Infrastructure Expansion

The $50M investment round coincides with a once-in-a-lifetime construction supercycle. Construction volume worldwide must grow significantly by 2030 to fill the void of digital infrastructure, address housing deficits in many parts of the world, and modify nearly every building in order to meet zero-carbon requirements. Data centre energy needs have skyrocketed, with high-density server racks increasing from standard 10 KWs to 100+ KWs, requiring dramatically new cooling solutions and complex electrical configurations.