Nuclideon Frequently asked questions
What is Nuclideon?
Nuclideon develops the Unlimited Detail engine, a patented CPU-native point cloud search system that renders datasets above 200TB at full resolution in real time, with no GPU required or used. The company was founded in August 2025 by the engineering team that built the engine, and holds the IP and patents outright.
What happened to Euclideon?
Euclideon commercialised the engine and went into administration in 2024 through commercial failure, not technology failure. Separately, Leica pulled out of its commercial partnership with Euclideon. Nuclideon is the original engineering team: it acquired the assets, cleared the debt, and now owns the engine outright. The engine never changed hands; the company around it did.
Do I need a GPU to render point clouds?
No. Instead of pushing geometry through a GPU graphics pipeline, the Unlimited Detail engine uses a search algorithm that finds only the points each pixel needs, running entirely on CPU. Render cost is fixed to screen resolution, not dataset size. The floor spec is a 4-core CPU with 8GB RAM and an SSD; integrated graphics handle display output only.
How large a point cloud can be rendered in real time?
The engine has demonstrated rendering above 200TB with no file-size ceiling, on a multi-core CPU with 8GB RAM. Render cost scales with screen resolution, not dataset size: the engine loads the points each pixel needs plus the index levels required to find them, so bandwidth and compute grow far more slowly than the data, not in proportion to it.
Can point cloud software run air-gapped or offline?
Yes. udStream can run fully offline in an air-gapped configuration: no telemetry, no licence call-home, local licence validation. That is a mode you enable, not the default, and the US Navy runs it that way on classified networks on standard government-issue hardware today. udServer brings the same browser-based streaming to an air-gapped network, deployed inside your own boundary.
What are udStream, udCloud and udSDK, and how do they relate?
One engine, three deployments. udStream is a desktop application that opens the full dataset on the hardware you carry. udCloud runs the same engine in the browser: rendering happens on the client CPU via WebAssembly, and only the points the current view needs travel over the network, so bandwidth tracks the screen and grows far more slowly than the dataset. There is no server-side rendering. udSDK embeds the engine inside your own platform as a compiled library. Pick the one that matches your environment, or combine them.
What file formats are supported?
Nuclideon ingests LAS, LAZ, E57, PLY, PTS, XYZ, and proprietary formats from many scanners including RIEGL. Datasets convert once to the UDS streaming format; after that they are never tiled, meshed, or re-processed again.
What hardware do I need?
For running udStream and converting datasets to UDS, the minimum is a 4-core CPU, 8GB RAM, and an SSD, with integrated graphics for display output only, no dedicated GPU. For larger datasets, multi-source fusion, or faster conversion, an 8-core CPU with 16GB RAM is recommended. udStream runs on Windows, macOS, and Linux. Viewing in udCloud needs only a modern browser.
How accurate are measurements?
Measurements run against every point the sensor captured, not a tiled or subsampled copy. Nuclideon does not introduce error at the rendering layer: a Leica ScanStation captured at plus or minus 3mm is measured at plus or minus 3mm. The accuracy your instrument delivers is the accuracy your deliverable reflects.
Does Nuclideon replace Esri, Hexagon or Trimble?
No. Nuclideon is not a replacement for your GIS, BIM, or asset-management tools. It solves the problem they cannot: rendering and streaming massive point clouds at full resolution on commodity hardware. udSDK embeds into your current systems; udStream and udCloud sit alongside them. It extends your existing workflows rather than competing with them.
How do I evaluate it?
Send us your hardest dataset. We render it at full resolution on standard hardware and walk you through the result. If we cannot, you owe us nothing. Evaluation runs on your data, your hardware, and your network, with a direct reply from the CEO or CTO.
How much does it cost?
udSDK is licensed per seat, from $150 per seat per year. There is a $25,000 minimum annual contract value across the whole agreement, not per seat, and per-seat rates fall at higher volume. The TCO calculator models the total against building in-house. udStream and udCloud pricing is published at launch. Contact us for the current pricing model.
Operate with all your Data, not a sample.
Operate with all your Data, not a sample.