Emmi AI
- Company typePrivate
- Founded2024
- HeadquartersLinz, Austria
- Headcount11–50
- GTM typeB2B
- OfferingSoftware
Emmi AI firmographics
Firmographics- Name
- Emmi AI
- Legal name
- Emmi AI GmbH
- Website
- https://emmi.ai
- Company type
- Private
- Founded year
- 2024
- Operating status
- Acquired
- Headcount range
- 11–50 employees
- Ownership category
- akta.pro rank
Emmi AI industry classification
Industry- Product category
- Industrial Engineering Simulation Software
- NAICS
- Computer Systems Design Services (541512), Custom Computer Programming Services (541511)
- SIC
- Services-Computer Integrated Systems Design (7373)
- akta.pro primary industry
- AI/ML Engineering & Model Development Services (BPAEAHAF)
Keywords
Where Emmi AI is headquartered
LocationHeadquarters
- HQ city
- Linz
- HQ country
- Austria
- HQ region
- Europe
Offices3 records
Markets served
Emmi AI business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Infrastructure, Marketing or Sales, Operations
Revenue model
- Enterprise Simulation Contracts: Custom AI simulation deployments to enterprise industrial customers with early contracts reaching 7-digit Annual Contract Values (ACV). Includes custom model training on proprietary datasets, enterprise support, SLAs, integration with existing simulation pipelines, and performance optimization for customer hardware.
- Commercial Noether Framework Licensing: Noether Framework is free for research and piloting under the Emmi Open Research License; commercial use requires a paid license. Enterprise support, SLAs, and custom model development offered as paid services.
- Professional Services & Custom Model Development: Dedicated training and workshops, integration assistance, performance optimization, and custom model development for enterprise customers and simulation software partners.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Other | Annual | Enterprise contracts — quote-based, 7-digit ACV reported for early customers |
| Freemium | Pay-as-you-go | Noether Framework — free for research/piloting; paid license for commercial use |
Go-to-market motion3 records
Distribution channels4 records
Marketing channels8 records
Emmi AI product offering
Product offeringCore offering
Emmi AI develops and sells Large Engineering Models (LEMs) — pre-trained transformer-based physics AI models that replace traditional numerical solvers for industrial engineering simulation. The company offers the open-source Noether deep learning framework, the AB-UPT (Anchored-Branched Universal Physics Transformer) architecture, and a portfolio of vertical-specific pre-trained models (NeuralWing for transonic 3D wing aerodynamics, NeuralMould for injection moulding, NeuralDEM for CFD-DEM multiphysics, GyroSwin for fusion plasma turbulence), sold to industrial enterprises with 7-digit ACV contracts and through CAE platform partners.
Product overview
Emmi AI is a deep-tech company building Large Engineering Models (LEMs) — foundation models for industrial physics simulation — and offers a platform-plus-modules architecture. At the core is the open-source Noether Framework, an industry-grade deep learning framework for Engineering AI that enables building, training, and deploying physics models. On top of the Noether Framework and its shared AB-UPT (Anchored-Branched Universal Physics Transformers) backbone, Emmi AI ships a portfolio of pre-trained, vertical-specific Large Engineering Models: NeuralWing (real-time transonic 3D wing aerodynamics), NeuralMould (injection moulding, co-developed with SIMCON as the Cadmould AI Solver), NeuralDEM (CFD-DEM particulate flow), and research models like GyroSwin (fusion plasma turbulence) and PAINT (neural digital twins). Together, these products target aerospace, automotive, semiconductor, and energy industries, replacing traditional numerical solvers with AI-driven real-time inference up to 1000x faster.
Differentiator
Problem solved
Functional benefit
Brands
- AB-UPT: Emmi AI's flagship foundation model for industrial physical simulation, released May 2025.
- NeuralMould
- Noether
- NeuralWing
Products and services
- Noether Framework Industry-grade open-source deep learning framework for Engineering AI that provides the full stack for training, evaluating, and deploying industrial physics models. Distributed as a Python library (pip install emmiai-noether) and on GitHub, with native integration to HuggingFace and AWS S3 for data and pre-trained model distribution. Free for research and piloting under the Emmi Open Research License; commercial use requires a paid license.
- NeuralWing Pre-trained AB-UPT Large Engineering Model for real-time neural simulation and design optimization of transonic 3D aircraft wing aerodynamics. Achieves 99.5% accuracy on pressure fields and 99.8% drag/lift data agreement; delivers 1000x speedup versus numerical RANS CFD with design-parameter optimization in 30 seconds. Built for aerospace engineering teams.
- NeuralMould First Large Engineering Model for injection moulding simulation. AB-UPT variant trained on almost petabytes of injection moulding simulations across thousands of materials. Delivers 200x speedup versus traditional solvers (1-2 seconds vs 30 min-1 hour) with consistent 5% relative errors. Supports any geometry, any material, and any injection gate.
- AB-UPT (Anchored-Branched Universal Physics Transformers) Emmi AI's flagship physics AI foundation model and core neural surrogate architecture for industrial physical simulation. Multi-branch Transformer that separates geometry, surface, and volume representations with anchored attention for linear-inference complexity on 100M+ cell meshes. Enforces physics consistency through a divergence-free vorticity formulation. First neural surrogate capable of end-to-end inference on 140M volumetric cells on a single NVIDIA GPU, directly from CAD geometry without remeshing.
- NeuralDEM First end-to-end deep learning surrogate for CFD-DEM multiphysics simulations of industrial particulate flows. Field-based physics representation with multi-branch neural operator coupling particle and fluid phases. Models systems with 500k particles and 160k CFD cells in real-time on a single GPU with mass conservation over long rollouts (28s, 2,800 ML timesteps). Now open source on HuggingFace and GitHub.
- GyroSwin Neural surrogate that simulates complex 5D gyrokinetic plasma turbulence in seconds rather than hours or days; up to 1000x faster than traditional computational methods. Built in collaboration with UK Atomic Energy Authority and JKU Linz; designed for next-generation fusion power plants such as the UK's Spherical Tokamak for Energy Production (STEP).
- Enterprise AI Simulation Deployment & Custom Model Development Custom AI simulation deployments offered to enterprise industrial customers as recurring subscriptions, including custom model training on proprietary datasets, enterprise support, SLAs, integration with existing simulation pipelines, performance optimization for customer hardware, dedicated training, and workshops.
Quantifiable outcome
- 1000x simulation speedup versus traditional numerical CFD solvers (across AB-UPT, NeuralWing, GyroSwin)
- +5 more outcomes
Companies that use Emmi AI
Customer profileNamed customers4 records
Segments4 records
Ideal customer profiles4 records
Emmi AI technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- No
- API docs
- API detail
Core technology
AI maturity
App detail
Integration6 records
AI capability10 records
Feature6 records
Emmi AI partnerships and signals
Strategic signalPartnerships
17 partnerships are on record, tiered flagship, minor and core.
- Mistral AIflagshipDefinitive agreement to acquire Emmi AI announced May 19, 2026 (closed/announced May 23, 2026). Emmi AI's 30+ researchers and engineers join Mistral AI's Science and Applied AI teams; Linz becomes an official Mistral AI office. The deal positions Mistral AI as the leading AI stack for industrial engineering across aerospace, automotive, and semiconductors. Johannes Brandstetter (Emmi AI's co-founder) becomes VP AI for Science at Mistral AI.
- PyTorch Foundation (Linux Foundation)minorEmmi AI joined the PyTorch Foundation as a new member (announced February 24, 2026), hosted by the Linux Foundation. Membership signals sustained engagement with the open-source agentic AI ecosystem and was announced ahead of PyTorch Conference Europe in Paris (April 7-8, 2026).
- SIMCONcoreCo-development partnership for the Cadmould AI Solver, the world's first Large Engineering Model for injection moulding simulation. Emmi AI trained NeuralMould on over a million transient trajectories across thousands of materials, partnering with SIMCON — a global leader in injection moulding CAE — to bring real-time AI simulation into injection moulding workflows.
- UK Atomic Energy Authority (UKAEA)minorCo-developed the GyroSwin AI model (with JKU Linz and Emmi AI) for 5D gyrokinetic plasma turbulence simulation. UKAEA plans to apply GyroSwin to next-generation fusion power plants including the UK's Spherical Tokamak for Energy Production (STEP).
- Johannes Kepler Universität Linz (JKU)coreAcademic/research partnership supporting the Noether Framework. Emmi AI is a spin-off from JKU Linz and NXAI; ongoing research collaboration with Prof. Dr. Sepp Hochreiter's institute. JKU is one of the named university endorsers of the Noether Framework.
- NXAIcoreEmmi AI was founded in December 2024 as a spin-off from JKU Linz and NXAI. CEO Albert Ortig of NXAI publicly endorsed the €15M seed round and the spin-out as a strong signal for Linz and Austria's position in industrial AI.
- Luminary CloudcoreSimulation platform partner supporting the Noether Framework. Emmi AI evaluated AB-UPT on Luminary Cloud's SHIFT-SUV (automotive) and SHIFT-Wing (aerospace) datasets, with results published in a Technical Report.
- RescalecoreSimulation platform partner supporting the Noether Framework for production-oriented engineering AI workflows.
- SimscalecoreSimulation platform partner supporting the Noether Framework for production-oriented engineering AI workflows.
- ETH ZürichcoreAcademic research group supporting the Noether Framework. ETH Zürich is named as one of the supporting university groups for the framework.
- University of WashingtonminorAcademic research group supporting the Noether Framework.
- Sorbonne UniversitéminorAcademic research group supporting the Noether Framework. Sorbonne is named as one of the supporting university groups.
- University of PennsylvaniaminorAcademic research group supporting the Noether Framework.
- Technische Universität München (TUM)minorAcademic research group supporting the Noether Framework. TUM is named as one of the supporting university groups.
- NVIDIAcoreNVIDIA is featured as a hardware partner/integrator on the Emmi AI website. Emmi AI's models (AB-UPT, NeuralWing, NeuralMould, NeuralDEM) are designed and benchmarked to run training and inference on single NVIDIA GPUs (including H100).
- Hugging FacecoreHugging Face is featured as an endorsing partner on the Noether Framework page; Emmi AI publishes models and datasets on Hugging Face (e.g., EmmiAI/NeuralDEM, EmmiAI/Emmi-Wing). The framework supports HuggingFace as a dataset storage provider.
- PhysicsXminorPhysicsX is featured as an endorsing partner of the Noether Framework on the Emmi AI website.
Scale indicators11 records
Recent moves7 records
Expansion highlights6 records
Emmi AI competitors and assessment
Company assessmentEmerging players
- Rescale: Cloud HPC platform for engineering simulation that is a Noether Framework partner. While Rescale is more of an infrastructure/orchestration layer, it increasingly embeds AI surrogates and overlaps in the AI-for-engineering workflow that Emmi targets.
- SimScale: Cloud-based CAE platform used by engineers for CFD/FEA/thermal simulation; named Noether Framework partner. Comparable industrial engineering end-customer base and an emerging interest in AI-augmented simulation, though broader in scope than Emmi's foundation-model approach.
- Luminary Cloud: Cloud-native AI-driven CFD platform for full-vehicle aerodynamic simulation; publishes the SHIFT-SUV and SHIFT-Wing datasets that Emmi benchmarks against. Overlap in AI-for-aerodynamics use cases, but Luminary is positioned as a simulation platform rather than a foundation-model provider.
Broad incumbents
- Altair Engineering: Engineering simulation and HPC software provider (Altair Inspire, HyperWorks, ultraFluidX) serving the same aerospace and automotive verticals as Emmi. Comparable physics-simulation scope with growing AI features but as part of a broad CAE portfolio.
- Ansys: Global incumbent in engineering simulation (Fluent, CFD, structural) with active AI/ML initiatives and an NVIDIA partnership. Represents the dominant incumbent whose installed base Emmi must displace or integrate with for industrial simulation workloads.
- Dassault Systèmes: 3DEXPERIENCE platform owner serving aerospace, automotive and industrial engineering with simulation, CAD and digital twin capabilities. Major incumbent whose CATIA/SIMULIA stack defines the design-to-simulation workflows Emmi is trying to accelerate.
- Siemens Digital Industries Software: Industrial software giant with Simcenter, Star-CCM+ and Xcelerator portfolio; embedded in the Emmi AI partner ecosystem as a customer/integrator logo. Comparable aerospace/automotive/energy customer base but a much broader portfolio.
Direct peers
- PhysicsX: AI-for-engineering startup building deep-learning surrogates and digital twins for industrial physics simulation (aerospace, automotive, energy). Direct overlap with Emmi's Large Engineering Model approach and target verticals; PhysicsX is even an endorsing partner of Emmi's Noether Framework.
- Neural Concept: Swiss ML startup providing AI-based surrogate models for engineering simulation (CFD, structural, thermal) used by automotive and aerospace OEMs. Closely comparable offering, customer base and go-to-market (enterprise engineering teams replacing traditional solvers with neural surrogates).
- Monolith AI: London-based AI platform for engineering R&D that combines physics-informed ML with experimental data to accelerate simulation and design. Comparable focus on industrial engineering customers and neural surrogate workflows.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat6 records
Key risks6 records
Key highlights7 records
Customer concentration
Emmi AI social profiles
Digital presenceEmmi AI compliance and trust
Trust signalCompliance2 records
Emmi AI financial estimates
Financial estimateRevenue estimate
Valuation estimate
Emmi AI leadership team
Management profileNumber of profiles
Profiles4 records
Emmi AI funding detail
Funding detailFunding overview
Funding rounds1 record
Investors4 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Emmi AI M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Emmi AI
What does Emmi AI do?
Emmi AI develops and sells Large Engineering Models (LEMs) — pre-trained transformer-based physics AI models that replace traditional numerical solvers for industrial engineering simulation. The company offers the open-source Noether deep learning framework, the AB-UPT (Anchored-Branched Universal Physics Transformer) architecture, and a portfolio of vertical-specific pre-trained models (NeuralWing for transonic 3D wing aerodynamics, NeuralMould for injection moulding, NeuralDEM for CFD-DEM multiphysics, GyroSwin for fusion plasma turbulence), sold to industrial enterprises with 7-digit ACV contracts and through CAE platform partners.
Is Emmi AI a public or private company?
Emmi AI is a private company. It is classified as corporate owned and is currently acquired.
When was Emmi AI founded?
Emmi AI was founded in 2024. It employs 11 to 50 people.
Where is Emmi AI based?
Emmi AI is headquartered in Linz, Austria, in the Europe region.
How does Emmi AI make money?
Three revenue lines are on record. Enterprise Simulation Contracts are the primary driver. The others are commercial Noether Framework Licensing and professional Services & Custom Model Development.
Who are Emmi AI's main competitors?
Emerging players on record are Rescale, SimScale and Luminary Cloud. Broad incumbents are Altair Engineering, Ansys, Dassault Systèmes and Siemens Digital Industries Software. Direct peers are PhysicsX, Neural Concept and Monolith AI.
Does Emmi AI have an API?
No public API is recorded for Emmi AI.
What industry is Emmi AI in?
Emmi AI's product category is Industrial Engineering Simulation Software. Its primary akta.pro industry code is BPAEAHAF, AI/ML Engineering & Model Development Services. Its NAICS code is 541512 and its SIC code is 7373.