Cambridge, MA
Heavy industry is paying an extraordinary price for operating without real-time intelligence. BP's 2010 Deepwater Horizon disaster — which killed 11 workers — ultimately cost the company $61.6 billion in total, including a $4 billion criminal fine, the largest criminal penalty against a single company in US history. More recently, in 2025, a Texas oil company executive was sentenced to prison and his firm fined $1.4 million after hydrogen sulfide deaths in the Permian Basin. In Canada, two companies paid over $550,000 in penalties following a fatal explosion at an Alberta oil and gas site. Regulators are no longer treating industrial safety failures as civil matters. They are criminal ones.
The financial damage from unplanned downtime is equally severe. An average offshore oil and gas site loses $38 million every year — roughly 27 days during which production halts completely while workforce costs, equipment leases, and site operations continue at full rate. Nothing coming in. Everything still going out.
Underlying both crises is a connectivity problem cloud computing cannot fix. Remote industrial sites generate terabytes of sensor and camera data daily, but with upstream bandwidth limited to a few megabytes per second and connectivity that is intermittent at best, transferring raw data to the cloud for analysis is simply not feasible. By the time data arrives and a decision returns, the window to act has closed. Data residency regulations make cloud dependency legally untenable too.
This is the problem MIT engineer Sidhant Kumar founded NZeroC AI to solve. The company's industrial edge AI platform processes data inside the facility — on ruggedised edge compute hardware integrated with Starlink — running advanced AI models, autonomous agents, and operational applications where data is generated. Real-time anomaly detection, predictive maintenance, workforce safety monitoring, and compliance automation, with raw data never leaving the premises.
NZeroC AI has received a grant from MIT Sandbox and is backed by the MIT Martin Trust Center and MIT Research Labs. It is already deployed with industrial organisations worldwide, reducing downtime, protecting workforces, and driving measurable improvements in profitability and compliance.
The companies that adopt intelligent edge AI will lower their costs, protect their people, and stay ahead of tightening regulation. The ones that don't will keep paying for it — in fines, in downtime, and in lives.