Tata Consultancy Services (TCS) has opened a Gemini Experience Center (GEC) at its Innovation Hub in Troy, Michigan, aimed specifically at accelerating AI-driven manufacturing solutions. Announced on March 9, 2026, the new facility is the seventh GEC in TCS’s network and the second in the United States; it was developed in partnership with Google Cloud.
What the Gemini Experience Center (GEC) is for, and what “Physical AI” means
TCS describes the Troy facility as a “Physical AI” GEC: a hands-on environment where manufacturers can explore, prototype and scale AI applications that operate in the physical, shop-floor world rather than only in data centers or back-office systems. The center bundles robotics (quadruped and humanoid), advanced sensing, edge intelligence and secure cloud orchestration into a TCS-branded framework called the TCS Physical AI Blueprint. Use cases the company highlights include autonomous patrolling and surveillance, environmental anomaly detection, PPE compliance monitoring, intelligent quality inspection, progress mapping and predictive equipment health monitoring.
TCS built the centre in collaboration with Google Cloud and says the GEC integrates Google’s Gemini models with TCS’s manufacturing engineering and systems integration capabilities. As part of a wider effort with Google Cloud, TCS also intends to broaden customer access to Gemini Enterprise and develop custom agents that can be integrated into industrial workflows. That combination, hyperscaler AI models plus systems integrator domain knowledge, is the model many large vendors are using to push AI into operational technology (OT) environments.
Anupam Singhal, President, Manufacturing, TCS, said, “Physical AI is where intelligence moves to the edge, into the real world of operations. With the launch of our Physical AI Gemini Experience Center for Manufacturing, we are enabling manufacturers to extend visibility and decision-making into environments that are difficult, risky, or inefficient for humans to access.”
An experience centre is useful for testing feasibility and for executive-level validation, but it is not a turnkey factory system. The demonstrations described (robotic patrolling, PPE monitoring, quality inspection) show technical capability; they do not guarantee immediate, low-cost deployment across diverse production lines. Manufacturers should expect further work on integration, cybersecurity, latency and scale-up engineering before these demos become reliable, long-running services on the shop floor. The center can shorten the time to insight, but converting that to operational benefit will still require tailored engineering and governance work.
If you are a manufacturing leader evaluating the Troy GEC or similar offerings, here are practical checks to run during any demonstration or pilot:
• Integration requirements: ask how sensors, PLCs and MES/ERP systems will connect, and for examples from lines with similar legacy equipment. (Inference based on the center’s end-to-end framing.)
• Data governance and security: confirm edge-to-cloud encryption, access controls and audit trails, because OT environments increase the attack surface.
• Resilience and latency: verify whether the proposed solution needs always-on connectivity or can operate autonomously at the edge during network outages.
• Measurable KPIs: require pilots to define and measure clear business metrics (safety incidents avoided, inspection yield improvement, mean time between failures reduction).
• Total cost of ownership: include integration, maintenance, model retraining and lifecycle upgrades in cost estimates rather than focusing on initial prototype cost alone. (This is an evidence-based recommendation inferred from standard industrial deployments.)
TCS says it will expand its Gemini Experience Center footprint to 13 centres worldwide by the end of 2026, positioning these GECs as nodes where enterprises, startups and universities can test and co-develop solutions. For the manufacturing sector, that signals increased vendor emphasis on “physical” AI use cases, a logical next step after a period when most industrial AI work focused on analytics and digital twins in the cloud. But success at scale will depend on how well vendors and manufacturers solve the messy engineering and organisational problems that follow a successful prototype.
The TCS Gemini Experience Center in Troy is a clear example of where vendors are directing investment: bringing large language and agentic AI models closer to the factory, and demonstrating robotic and sensing use cases that are hard to simulate. For manufacturing leaders, the centre is a useful place to gain technical clarity and to test real scenarios under guided conditions. For tangible operational value, however, companies must treat demonstrations as the start, not the end, of the work: plan for integration, governance, and long-term product engineering if they want pilots to mature into reliable, cost-effective production systems.
