Inspiration
Cascade came from a simple question we kept asking ourselves as we learned about industrial IoT: Why do factories still get blindsided by failures even when they have so many sensors and monitoring tools?
As we dug deeper, we discovered something surprising. Most predictive maintenance systems do a good job of telling you when one machine is about to fail—maybe a pump is overheating or a motor is vibrating more than usual. But real factories don’t operate one machine at a time. They run as connected systems, where machines depend on one another like links in a chain.
And that’s where the real problem appears. A small issue in one machine can quietly push stress into the next one, and before anyone realizes it, an entire production line is down. We found that 73% of unplanned downtime comes not from the first failure, but from the failures that follow it. That statistic completely changed our understanding of industrial operations.
We also learned that operators usually get only 15–30 minutes of warning before something critical happens, far too little time to gather maintenance staff, find spare parts, or safely shut things down. They actually need 2-4 hours, but the system doesn’t give them that window. So even when the alarms work, it’s already too late.
That’s when we saw a clear gap: No one is predicting the chain reaction. Everyone is predicting the first failure, but the first failure isn’t what hurts the most.
This idea connected beautifully with what Toshiba and Fixstars specialize in. Toshiba works on digital transformation, IoT, and building more resilient operations. Fixstars focuses on high-performance computing and optimization, exactly the kind of intelligence needed to evaluate multiple scenarios and choose the best action.
It felt like the perfect opportunity to ask: What if we could see the entire failure cascade before it starts? What if we could understand the actual impact, and not just the first alarm? And what if we could recommend the best intervention strategy automatically?
That’s how Cascade was born, a platform built not just to predict a problem, but to understand how that problem could spread and how to stop it from turning into a costly system-wide failure.