Rewiring the Blueprint: How AI is Breaking the MEP Engineering Bottleneck

LAS VEGAS, NV – September 16, 2026 — Every major building project on earth eventually collides with the exact same invisible ceiling: the human engineering capacity required to make it a reality. As the global economy races to construct hyperscale data centers, electrify aging infrastructure, and meet stringent net-zero mandates, the mechanical, electrical, and plumbing (MEP) engineering sector has become the primary bottleneck of the modern built environment.

Today, Stockholm-based AI developer Endra aims to shatter that ceiling. At its Epoch launch event in Las Vegas, the company unveiled Endra Power Studio, an agentic AI platform purpose-built to automate the most labor-intensive aspects of electrical engineering. Concurrently, Endra announced the acquisition of Swiss AI laboratory Planlabs, a strategic move designed to bring sophisticated mechanical engineering capabilities in-house.

Propelled by a recently closed $50 million Series A funding round led by Andreessen Horowitz—with participation from Notion Capital and Norrsken VC—the platform's debut signals a pivotal shift in a $150 billion global services market historically constrained by the sheer number of available engineers.

Breaking the Data Center Bottleneck

The commercial traction of agentic platforms like Endra is being driven by a convergence of macro-economic pressures. In North America alone, data center capacity under construction has surpassed 66 gigawatts. Within these massive hyperscale facilities, MEP design dictates upwards of 70 percent of total construction complexity and cost. Yet, developers consistently report that engineering design turnaround is now their primary limiting factor, outpacing even the availability of silicon.

Coupled with a demographic wave of retiring senior engineers and new decarbonization mandates that multiply the mathematical calculations required for every square foot designed, the industry is starved for capacity.

Endra Power Studio attempts to solve this equation by replacing a fragmented ecosystem of disconnected tools with a unified, end-to-end electrical design platform. Traditionally, engineers have been forced to act as human APIs—drafting in Autodesk's Revit, calculating loads in Excel, and coordinating spatial clashes in Navisworks.

With Power Studio, engineers load an architectural model, establish design requirements, and instruct AI agents to execute the workflow. The company claims that a fully code-compliant electrical design for a 500,000-square-foot commercial building—a process that typically consumes two months of manual drafting and calculation—can now be completed in less than a day.

This radical compression is achieved through features like "Type Rooms," which allow engineers to design a standardized space, such as a hospital exam room or hotel suite, and programmatically propagate that design across hundreds of identical geometries in the model. Additionally, firms can encode their specific engineering standards, unit placement rules, and code interpretations into "Playbooks," ensuring the AI adheres strictly to the consultancy's proprietary design philosophy.

While internal case studies boast efficiency multiples of up to 70x on isolated tasks like fire alarm routing, practicing MEP principals note that real-world deployment requires caution. Automated building design relies heavily on clean architectural inputs. If an upstream architectural Building Information Model (BIM) features unclosed boundary spaces or incomplete ceiling grids—a frequent reality in the industry—the automated routing must adapt or pause for human intervention.

Beyond Revit: Challenging the Legacy BIM Paradigm

For over two decades, legacy software giants like Autodesk, Bentley Systems, and Trimble have dominated the architecture, engineering, and construction (AEC) software landscape. Endra's approach is distinctly collaborative rather than adversarial, positioning itself as an intelligence layer that sits directly above existing systems. The platform ingests and outputs native Revit models, single-line diagrams, and panelboard schedules, ensuring documentation remains tightly coupled to the design without disrupting established file formats.

However, the introduction of agentic AI introduces a fascinating structural conflict for global engineering consultancies. Enterprise partners currently piloting or utilizing Endra—including heavyweights like AtkinsRéalis, Ramboll, Buro Happold, Hoare Lea, and AFRY—traditionally operate on a Time-and-Materials (T&M) or billable-hour business model.

If a technology compresses hundreds of billable hours into a single afternoon, the traditional revenue model breaks down. Industry analysts suggest this will accelerate a shift toward value-based pricing and fixed-fee deliverables. Under fixed-fee structures, the efficiency gains delivered by platforms like Power Studio accrue directly to the consultancy's gross margins. Furthermore, by multiplying the output of their existing workforce, perpetually understaffed firms can finally accept backlogged project bids without needing to aggressively expand headcount.

The competitive landscape is already reacting. In late 2025, engineering giant AECOM acquired Consigli to automate mechanical room planning, while startups like Augmenta have raised significant venture capital to tackle generative MEP routing. As these agentic platforms scale, legacy incumbents face mounting pressure to innovate beyond the limitations of decades-old desktop codebases.

The Physics Problem: Why Buildings Need More Than Language Models

Unlike generating text or writing software code, engineering physical infrastructure requires strict adherence to the laws of physics, spatial three-dimensional topology, and rigid building codes. In the built environment, a large language model (LLM) "hallucination"—such as routing an HVAC duct directly through a load-bearing structural I-beam—can result in millions of dollars in field change orders.

This fundamental "physics problem" explains the strategic necessity behind Endra's acquisition of Planlabs. Founded in Aarau, Switzerland, in June 2024, Planlabs is driven by a team of five PhDs specializing in mathematics and computer science. Rather than relying on probabilistic text prediction, the startup developed deterministic, physics-based constraint solvers and computational geometry algorithms specifically for mechanical systems.

"We started Planlabs with the belief that mechanical systems can be engineered fundamentally differently, automatically generating coordinated, collision-free solutions while keeping the engineer in control," said Rea Sodero, Co-Founder and CEO of Planlabs. "With Endra, we have found a team that shares this vision and has the ambition, technology and reach to bring it to engineers around the world. Together, we have the opportunity to build something much bigger: a platform that transforms how every MEP discipline is engineered."

The integration of Planlabs' technology serves as the foundation for Endra's mechanical engineering roadmap. While Power Studio currently handles electrical workflows, a mechanical and HVAC module is slated to follow once the Swiss team's computational geometry engine is fully integrated. A plumbing module is scheduled for 2027, which will complete Endra's vision of a unified, tri-discipline MEP platform.

Reclaiming Engineering Judgment

Despite the rapid advancement of generative design, the human element remains legally and practically indispensable. In highly regulated markets across North America and Europe, electrical and mechanical designs must be rigorously reviewed, stamped, and sealed by licensed Professional Engineers or Chartered Engineers. Building authorities do not accept automated AI submissions.

Recognizing this, Endra's architecture is fundamentally built around a human-in-the-loop philosophy. The AI stages its outputs, but the engineer sets the initial requirements and signs off at every critical juncture. Furthermore, to satisfy enterprise security demands, the platform is SOC 2 Type II and ISO 27001 certified, and strictly prohibits the training of public models on proprietary customer data.

"Endra takes care of the repetitive work, and gives the engineers back the time for judgment and design," said Niklas Lindgren, CEO and Co-Founder of Endra. "A firm's engineering judgment is its product, and Playbooks let a firm hold that as a standard its whole team designs against, with the engineer setting the requirements and signing off at every stage."

As the global construction industry grapples with the unprecedented demands of the AI era and the green energy transition, the automation of repetitive drafting represents more than just a software upgrade. It is a necessary evolution, ensuring that the world's most critical infrastructure can actually be built before the talent required to design it ages out of the workforce.