Sona Comstar commits Rs 62 crore, gives no AI timeline

Sona Comstar has allocated Rs 62 crore of capital expenditure to robotics and physical artificial intelligence, while saying the field represents 4% of its Rs 25,000 crore order book. Sonacom has orders and is deploying autonomous mobile robots on factory floors, but management declined to forecast when the business could become sizeable.

How much is Sonacom spending on robotics and physical AI?

Sonacom has allocated Rs 62 crore of capital expenditure, or capex, to robotics and physical AI, and management said that amount should cover the planned investment in the area. The allocation was disclosed in the previous quarter, according to management, while research and development work is being conducted in-house by Novelic and driveline teams.

Management did not provide a separate operating research and development budget for robotics, physical AI, humanoids or autonomous mobile robots. Instead, it said a substantial part of the effort is employee time: Novelic teams and driveline teams are carrying out their regular roles while devoting time and management bandwidth to technology development. The Rs 62 crore figure is therefore the disclosed capex measure, not a total cost estimate for all internal development work.

Sonacom said robotics and physical AI account for 4% of its Rs 25,000 crore order book, which management characterized as small relative to the company’s current scale. It said the area will grow, but did not disclose the value of individual robotics orders, the timing of their revenue conversion, or capex split by product category.

What robotics products and markets is Sonacom targeting?

Sonacom says it has capabilities relevant to 60% of the bill of materials, or BOM, for humanoid robots, including hardware and perception software. Management presented humanoids as one form factor within a wider physical-AI opportunity, rather than as the company’s only target product or market.

The company identified autonomous mobile robots, or AMRs, intelligent industrial robots and robotaxis as other application areas. Management described robotaxis as physical-AI deployments that must operate at speed and in varied real-world environments, making their task more complex than many fixed-location applications. Sonacom said it aims to be a global supplier, but did not name customers or give product-validation dates.

The company also said it has capability in the perception stack, the software layer used by intelligent machines to interpret their surroundings. That claim extends Sonacom’s stated addressable content beyond mechanical and driveline hardware, although management did not quantify its expected content value per robot, robotaxi or industrial system.

Why is Sonacom not predicting a robotics inflection point?

Sonacom is not predicting an inflection year because management said technology-adoption curves are difficult to time. It compared the uncertainty with its electric-vehicle business, which management said represented 1% of revenue in 2019 but now accounts for 51% of revenue after several earlier years without material growth.

Management said the eventual growth curve could be steep, but did not provide a year, revenue target or order-win target for robotics and physical AI. The EV comparison was offered as context for why management does not want to attach a public forecast to a market that may remain small for longer than initial expectations suggest.

On the cost side, management cited open-source physical-AI models and a roughly 1,000-times reduction in compute costs. It said the constraint is not moving as quickly from simulation to deployment in real-world environments, where machines need reliable training and performance across actual operating conditions. That unresolved simulation-to-reality gap is central to management’s refusal to set a commercial-scale timetable.

How is Sonacom using AMRs in its factories?

Sonacom is deploying AMRs on its factory floors to develop use cases, obtain real-work training and use feedback to improve the products. Management said these deployments are occurring as it spoke on the call, making AMRs the disclosed live internal application of its physical-AI work.

The company did not disclose the number of AMRs, the factory locations, the tasks being performed or any productivity impact. Its stated purpose is a product-development feedback loop: factory deployment provides operating data and use cases that can improve AMR design, rather than relying only on simulation or presentation material.

This internal work sits alongside the 4% robotics and physical-AI share of Sonacom’s Rs 25,000 crore order book. Management’s comments distinguish between having orders and reaching material scale: it has cited current orders and current factory deployment, while withholding figures on customer rollout volumes and revenue timing.

Will Sonacom deploy humanoid robots on production floors?

Sonacom is evaluating humanoid robots for production-floor use, but management said deployment will depend on whether the machines meet normal payback and return tests. It said humanoids should be assessed as capital equipment in the same manner as industrial robots or other machinery.

Management identified unsafe tasks or work with low acceptability for people as potential early use cases. It said replacing people solely for the sake of replacement is not its objective, and that any decision must make business sense. The company did not announce a humanoid purchase, factory trial, deployment date or budget beyond the Rs 62 crore capex allocation for the broader robotics and physical-AI area.

Sonacom separately said AMRs are already being deployed on factory floors, while humanoids remain under evaluation. That difference matters because AMRs are part of the company’s current real-work training programme, whereas humanoid use remains conditional on task suitability, payback and returns.

Conclusion

Sonacom’s disclosed strategy combines Rs 62 crore of capex, a 4% share of its Rs 25,000 crore order book and internal AMR deployment to develop real-world applications. It says its capabilities cover hardware and perception software across humanoids, industrial robots and robotaxis, but management has not equated those capabilities or orders with a forecast of commercial scale.

The next disclosed event is Bharat Expo in February, where management said it expects to demonstrate technologies under development, including the work discussed during the call. The key issues to watch are whether factory use cases improve AMR products, whether new deployments validate those products in real-world settings, and whether the 4% order-book contribution expands.