Uploaded October 2026 | Updated November 2026, 2 days ago
In this interview from theCUBE + NYSE Wired: Physical AI & Robotics, Igor Marinelli, chief executive officer of Tractian, joins theCUBE + NYSE Wired's Gemma Allen to discuss how industrial AI is teaching machines to signal failure before it happens. Marinelli traces Tractian's origin to his father's 35 years as a maintenance technician at a paper mill, a background that exposed him early to the $1.4 trillion annual cost of unplanned downtime. He explains why factories remain far from "smart," noting that the average industrial asset is 24 years old and many operations still depend on manual inspections in hazardous environments.
The conversation also explores why OEMs have little incentive to modernize maintenance, since aftermarket parts and reactive service contracts remain lucrative revenue streams. Marinelli details how Tractian's asset retrofitting approach levels instrumentation across decades-old equipment, delivering remaining-useful-lifetime data that frees customers from calendar-based schedules and OEM lock-in on parts. He highlights the company's move toward domain-specific AI, post-training models on open-source foundations like Nemotron to keep customer data secure while outperforming generalist systems. With a 90-day average deployment cycle and new offices in Poland and Singapore to follow global customers, Marinelli outlines how Tractian is scaling predictive maintenance into a new standard for industrial reliability.
Find more SiliconANGLE news and analysis siliconangle.com
Follow theCUBE's wall-to-wall event coverage siliconangle.com/events
Learn about the latest theCUBE events thecube.net
00:00 - Industrial AI: Predictive Maintenance and Adoption Barriers
03:56 - Smarter Maintenance: Incentives, Context, and Parts
10:13 - Asset Reliability and Supply Chain Continuity
14:53 - Deploying and Scaling Secure Domain-Specific AI Globally
#theCUBE #PhysicalAIRobotics #theCUBEresearch #NYSEWired #Tractian #PredictiveMaintenance #IndustrialAI
In this interview from theCUBE + NYSE Wired: Physical AI & Robotics, Igor Marinelli, chief executive officer of Tractian, joins theCUBE + NYSE Wired's Gemma Allen to discuss how industrial AI is teaching machines to signal failure before it happens. Marinelli traces Tractian's origin to his father's 35 years as a maintenance technician at a paper mill, a background that exposed him early to the $1.4 trillion annual cost of unplanned downtime. He explains why factories remain far from "smart," noting that the average industrial asset is 24 years old and many operations still depend on manual inspections in hazardous environments.
The conversation also explores why OEMs have little incentive to modernize maintenance, since aftermarket parts and reactive service contracts remain lucrative revenue streams. Marinelli details how Tractian's asset retrofitting approach levels instrumentation across decades-old equipment, delivering remaining-useful-lifetime data that frees customers from calendar-based schedules and OEM lock-in on parts. He highlights the company's move toward domain-specific AI, post-training models on open-source foundations like Nemotron to keep customer data secure while outperforming generalist systems. With a 90-day average deployment cycle and new offices in Poland and Singapore to follow global customers, Marinelli outlines how Tractian is scaling predictive maintenance into a new standard for industrial reliability.
Find more SiliconANGLE news and analysis siliconangle.com
Follow theCUBE's wall-to-wall event coverage siliconangle.com/events
Learn about the latest theCUBE events thecube.net
00:00 - Industrial AI: Predictive Maintenance and Adoption Barriers
03:56 - Smarter Maintenance: Incentives, Context, and Parts
10:13 - Asset Reliability and Supply Chain Continuity
14:53 - Deploying and Scaling Secure Domain-Specific AI Globally
#theCUBE #PhysicalAIRobotics #theCUBEresearch #NYSEWired #Tractian #PredictiveMaintenance #IndustrialAI






