2025: A Year of Momentum and Looking Ahead to 2026

As 2025 comes to a close, I find myself reflecting on the progress we’ve made since my co-founder, Lars Tingelstad, and I first committed to this vision in 2024. Our belief was simple: the future of robots depends on skills created from the union of data driven models and optimal control. For the manufacturing world, we believe this union is the best way to achieve a level of automation that can expand the potential of robotics to meet today’s labor and production demands. Today, we are seeing that vision start to become a reality.

Some of the major milestones from this past year include:

Validation from the Manufacturing Industry

T-ROBOTICS won the prestigious CECIMO Machine Tools Innovation (MTI) Award at EMO Hannover 2025. This was a significant honor and a major step for us, as it was our first time attending the world's leading trade fair for production technology with our own booth. To illustrate the flexibility of our solutions, we featured two interactive demos: CNC Milling Automation with ABB, showcasing hyper-flexible vision-based part handling, and CNC Turning Automation with Universal Robots, highlighting intelligent job orchestration. Both demos allowed visitors to define machining jobs conversationally using ActGPT, proving that AI-powered, adaptive machine tending can eliminate rigid fixturing and enable true part variability on the shop floor.

ActGPT: Delivering Impact in Production

This year, we brought ActGPT from an idea to production. The difference between how our customers used robots a year ago and how they use them now is significant. We have moved away from procedural programming to a system where a machinist can handle changeovers and configure the robot for new batches in minutes using conversational instructions. Behind the scenes is one of the most sophisticated tech stacks in the world, including a VLA transformer tuned for machining actions. We did not just create a model; we created a solution that delivers one of the shortest return on investment in the machining industry.

Global Growth and Operations

The exceptional momentum of T-ROBOTICS is solely because of the quality of people that make our company. We finished 2025 with 25 employees, having hired key talent across all teams including commercial, product, and software engineering. To support this growth, we deployed the seed capital raised at the end of 2024, led by Emergent Ventures and Engine Ventures. This allowed us to open our headquarters in Silicon Valley and Trondheim, Norway. With this world-class team and our unique depth in pre-training robot skills, we look forward to automating CNC machining and beyond for our customers.

A Year of Strategic Collaboration

We are not in this alone; our success is tied to an ecosystem dedicated to Physical AI. Our software is now compatible with robots from Universal Robots, ABB, and FANUC, as demonstrated through our partnership with ABB to power the OmniVance™ Collaborative Machine Tending (OCMT) cell. Central to our scale are integrators like SE Automation, Kameleon Robotics, and Fluidotronica, with whom we work to deploy our technology and deliver maximum value to customers. This collaborative approach, showcased this year at NVIDIA GTC and Hannover Messe, is key to making automation accessible worldwide. Looking ahead, we believe that five years from now, most robots will be powered by pre-trained AI models, and we intend to work with OEMs to make that future a reality. We are excited about the many new and expanded partnerships coming in the year ahead.

Technical Evolution: Addressing Market Trends in 2026

As we move into the new year, one reality is clear. Robots are limited in their performance, while the expectations of the manufacturing world and its labor needs demand that robots do more. To close this gap, we are focusing on the key technical trends that will define the next generation of industrial robotics. 

The Shift to Highly Specialized Models

While general AI captured headlines in the past, 2026 will shift to specialized models. Our work will focus on these highly specialized, compact, pre-trained models that are fully deployable on edge for all types of industrial tasks. These models are tuned specifically for the manufacturing environment, ensuring they provide the intelligence the industrial world demands without the overhead of generalized systems.

The Move Toward Agentic Autonomy 

The industry is moving from programmed motions to agentic behavior where robots possess more intelligence. We are focusing our efforts on the agent and the training of VLA models. By prioritizing world representation and modeling, we enable our robots to understand their surroundings and execute plans that adapt dynamically to sensor feedback. This gives unprecedented intelligence to industrial robots.

Simulation-Driven Intelligence and Failure Modeling

The most significant bottleneck in AI is high-quality data. In 2026, the leaders in robotics will be those who can leverage real and synthetic data to make deployable AI models in record time. We are focusing on data pipelines including highly physics accurate synthetic data. 

The opportunity in physical AI is massive, and we need to fill it. We have the team, the technical expertise, and the partnerships to make it happen. We have some very exciting announcements coming in the new year, and I look forward to sharing more with you soon. 

Onwards to 2026.