# Skild AI Hits $100M ARR in 10 Months Proving Physical AI Enterprise Value

> The robotics software firm scaled from initial deployments to more than 60 paying customers, placing hundreds of robots in manufacturing, logistics, food service and data centers while forming partnerships with NVIDIA, Foxconn and others.

*Published 2026-09-11 · By The Intel Desk*

Skild AI is a robotics software company that develops general-purpose AI systems allowing robots to learn tasks from video demonstrations without weight updates or task-specific training.

Skild AI has crossed the threshold of $100 million in annual recurring revenue run rate in a remarkably short period of ten months following its initial commercial deployment. This milestone is significant because it shows the speed at which physical AI technologies can be adopted in real world enterprise settings. The company has not only achieved this revenue but has also expanded its customer base to more than sixty paying entities. These customers are utilizing hundreds of robots in a variety of applications ranging from manufacturing and logistics to inspection and security. Additionally, the robots are being used in food preparation and within warehouses, factories, and data centers. The technology behind this success is the S1 model which allows for in context learning from video. This means that robots can perform tasks they have not been specifically trained for by observing a single demonstration. The partnership with NVIDIA and Foxconn is facilitating the use of dual arm manipulators for precise assembly work on advanced computing systems. Other pilots are underway with Mitsui and Company in commercial kitchens and with Sumitomo Wiring Systems in wire harness production. The revenue in the previous year was around thirty million dollars indicating the explosive growth in the recent period. This development is changing how enterprises approach automation by reducing the need for extensive programming and custom development for each new task.

## What background led to the shift from robotics demos to paid deployments?

In the past, robotics systems in enterprise environments were often limited to repetitive tasks that required extensive preprogramming and customization for each specific application. This approach made it difficult and costly to deploy robots in dynamic or varied settings. Skild AI's technology addresses this limitation by enabling robots to learn from experience through video demonstrations. The co founder and chief executive officer Deepak Pathak has emphasized that learning by experience rather than preprogramming represents a major step change in the field of robotics. This shift is allowing companies to deploy robots more quickly and with greater flexibility. The background of the company includes collaboration with major players in the technology and manufacturing sectors. The involvement of entities such as NVIDIA provides access to advanced hardware that supports the computational needs of the AI models. Foxconn brings expertise in large scale manufacturing which is essential for scaling the deployments. The overall context is one of increasing interest in physical AI as a way to address labor shortages and improve efficiency in industrial operations. The rapid revenue growth is a testament to the market demand for such solutions that can be implemented without long development cycles. Additional partners listed among the entities include ABB Robotics, Universal Robots, SoftBank and Zebra Technologies though specific deployment details for these firms remain part of broader ecosystem expansion.

The enterprise AI landscape has long sought methods to make robotic systems more adaptable without incurring high costs for each new task. Traditional methods involved lengthy training periods and hardware specific programming that limited scalability. Skild AI's model bypasses many of these barriers by relying on in context learning. This allows the system to generalize from a single video to perform tasks lasting up to ten minutes that were previously unseen. The result is a reduction in the time from demonstration to operational deployment. Enterprises in sectors such as manufacturing and logistics benefit from this flexibility because production lines and warehouse operations often require quick adjustments to new product variants or process changes. The background also includes the recognition that data centers and construction sites present unique challenges that benefit from general purpose capabilities. Security and inspection tasks further expand the addressable market. The transition documented in industry reports shows that the era of limited demos has given way to sustained commercial use across multiple verticals.

## What technical specifics define the S1 model's in-context learning approach?

The S1 model from Skild AI operates by processing a single video demonstration to acquire the ability to execute new tasks. This process occurs through in context learning which does not involve updating the underlying model weights or performing task specific post training. Tasks can last up to ten minutes and include complex sequences that were not part of the original training data. The approach enables robots to handle variability in real world environments such as differences in object positions or lighting conditions. In enterprise settings this capability translates to faster integration because operators can provide a demonstration rather than requiring software engineers to code each variation. The technical foundation supports dual arm manipulators in partnership deployments where high precision is required. Assembly of NVIDIA Blackwell systems exemplifies the level of accuracy achieved. The model supports operations in commercial kitchens where food preparation involves handling diverse ingredients and tools. Wire harness manufacturing requires precise manipulation of cables and connectors. These technical attributes reduce the barrier to entry for enterprises seeking to automate previously manual processes.

Further technical details indicate that the learning mechanism allows generalization across previously unseen scenarios. This stands in contrast to earlier robotic systems that required exhaustive data collection and retraining for each new environment. The in context approach preserves the base model while adapting behavior on the fly. This design choice supports deployment at scale because the same model can serve multiple customers with different task requirements. Hundreds of robots operating across sixty plus sites demonstrate the robustness of the underlying architecture. The system handles moving goods in warehouses, making deliveries in logistics operations, inspecting sites for quality control, and providing security through monitoring functions. Each application benefits from the ability to learn from minimal input. The technical specifics also align with hardware platforms from partners that supply the necessary compute and sensor integration. This combination of software intelligence and hardware capability forms the basis for the observed revenue acceleration.

## Which partnerships are expanding Skild AI deployments in manufacturing and food service?

Partnerships play a central role in Skild AI's expansion strategy. Collaboration with NVIDIA and Foxconn focuses on deploying the AI brain on dual arm manipulators for high precision assembly of NVIDIA Blackwell systems. This work occurs in manufacturing environments where accuracy and repeatability are critical. The partnership combines software learning capabilities with hardware platforms suited for complex assembly lines. Additional pilots include commercial kitchen operations with Mitsui and Company where robots perform food preparation tasks. Wire harness manufacturing with Sumitomo Wiring Systems tests the system in environments requiring fine motor control and handling of flexible materials. These partnerships illustrate how the technology integrates into established industrial workflows. The deployments span multiple sectors and demonstrate the general purpose nature of the S1 model. Enterprise customers gain access to automation without investing in bespoke development for each new task. The partnerships also provide validation that supports further customer acquisition.

Key Skild AI deployment sectors and partnersSectorPartnersTasksManufacturingNVIDIA, FoxconnHigh-precision assembly of Blackwell systemsFood PreparationMitsui & Co.Commercial kitchen operationsWire Harness ManufacturingSumitomo Wiring SystemsCable and connector assemblyLogistics and WarehousesMultiple enterprise customersMoving goods and deliveriesInspection and SecurityMultiple enterprise customersSite inspection and monitoring

## What market and stakeholder implications arise from this deployment flywheel?

The market implications of Skild AI reaching one hundred million dollars in annual recurring revenue run rate within ten months extend across the enterprise AI and robotics sectors. Stakeholders in manufacturing and logistics now have evidence that general purpose robotic systems can generate measurable returns without prolonged customization. This reduces perceived risk for companies considering automation investments. The scaling to sixty plus customers with hundreds of robots indicates that the technology addresses real operational pain points such as labor availability and process variability. Food service and data center operators gain similar options for tasks that were previously difficult to automate. The revenue trajectory from approximately thirty million dollars in twenty twenty five to the current run rate suggests accelerating adoption. Stakeholders including investors and potential partners can view this as validation of the physical AI approach. The implications also include competitive pressure on traditional robotics providers that rely on task specific programming. Enterprises may shift procurement strategies toward solutions that offer rapid deployment and adaptability.

Further market analysis shows that the deployment flywheel creates momentum for additional use cases. As more robots operate in the field the volume of demonstration data increases which in turn supports broader generalization. This positive feedback loop benefits both the technology provider and its customers. Manufacturing stakeholders can achieve higher utilization of capital equipment by automating variable tasks. Logistics operators improve throughput in warehouses where demand fluctuates. The presence of partners such as ABB Robotics, Universal Robots, SoftBank and Zebra Technologies in the broader ecosystem suggests potential for integrated solutions that combine multiple hardware and software offerings. The overall effect is a maturation of the physical AI market that moves beyond proof of concept projects to sustained commercial operations. This shift has implications for workforce planning as enterprises reallocate human resources from repetitive tasks to higher value activities.

> This is a major milestone for us as it signifies a new chapter for robotics, moving from an era of demos to an era of deploymentsDeepak Pathak, co-founder and Chief Executive Officer

## What expert reactions underscore the transition in robotics?

Expert commentary on the achievement emphasizes the fundamental change in how robotics systems are developed and deployed. Deepak Pathak has stated that learning by experience and not preprogramming is the step change that has happened in robotics. This perspective aligns with the observed revenue growth and customer expansion. Industry observers note that the ability to learn tasks from a single video demonstration addresses a long standing limitation in the field. The reactions highlight that enterprises are now willing to pay for solutions that deliver immediate operational value. The transition from demos to deployments is presented as evidence that physical AI has reached a level of maturity suitable for production environments. Additional reactions from partnership announcements focus on the precision achieved in assembly tasks and the flexibility demonstrated in kitchen and manufacturing pilots. These comments collectively reinforce the narrative that the technology is ready for broader enterprise adoption.

## What is next for Skild AI and physical AI in enterprise settings?

Looking ahead the company is positioned to expand its customer base and deployment volume further. The current scale of sixty plus customers and hundreds of robots provides a foundation for additional growth in existing sectors and entry into new ones. Continued partnership development with hardware providers will support larger and more complex installations. The technical capabilities of the S1 model suggest that future enhancements could extend task duration and complexity while maintaining the single demonstration learning approach. Enterprise stakeholders can expect increased availability of robots for inspection, security and logistics applications. The market implications include potential standardization of in context learning methods across the industry. The next phase will likely involve deeper integration with existing enterprise software systems for scheduling and monitoring. This evolution will further embed physical AI into daily operations across manufacturing, food service and data center environments.

- Initial commercial deployment established the foundation for revenue generation.
- Expansion to sixty plus paying customers demonstrated market acceptance across sectors.
- Achievement of one hundred million dollar annual recurring revenue run rate validated the business model.
- Partnerships with NVIDIA, Foxconn, Mitsui and Company and Sumitomo Wiring Systems enabled specialized deployments.
- Pilots in kitchens and wire harness manufacturing opened additional vertical markets.
- Future scaling expected through broader task generalization and hardware integrations.

## Sources

1. [Skild AI crossed $100 million in annual revenue run rate ten months after first commercial deployment and scaled to 60+ paying customers with hundreds of robots.](https://www.skild.ai/blogs/skild-crosses-100m-arr)
2. [Skild AI reached $100 million in recurring revenue run-rate 10 months after beginning commercial use.](https://www.bloomberg.com/news/articles/2026-09-10/robotics-startup-skild-ai-hits-100-million-in-revenue-run-rate)
3. [The company reached a $100 million annual revenue run rate 10 months after its first commercial deployment and has built more than 60 deployment partnerships.](https://blogs.nvidia.com/blog/skild-ai-s1-physical-ai/)

---
Source: https://aiintelreport.com/enterprise-ai/skild-ai-100m-arr-enterprise-ai-deployments
Index: https://aiintelreport.com/llms.txt · Full text: https://aiintelreport.com/llms-full.txt