Funding for robotics has cooled overall since its 2021-2022 peak, but many of the issues exposed by the pandemic remain persistent. The biggest driver of venture funding in the sector is the ongoing labor shortage. Analyst firm Garner predicts that half of large companies will have robots in their warehouses or manufacturing processes by 2028.
Another key factor in favor of warehouse/logistics robots is that they have a proven track record: While many approaches to automation today have a theoretical ROI, many companies, including Amazon, are already using warehouse robots in the wild.
GrayMatter is one company with a proven track record in this field: the Southern California company self-reports that its system now “improves production line productivity by 2-4 times.” [and a] “Reduced consumable waste by more than 30%.” Currently, major companies such as Boeing and 3M are using the company's system.
This is despite GrayMatter being a young company that was only founded towards the start of the 2020 pandemic.
“We founded GrayMatter to increase productivity while prioritizing employee well-being,” co-founder and CEO Aryan Kabir said in the release. “With our physics-based, AI-powered systems, we are unlocking new levels of efficiency and productivity while delivering on our mission. With the backing of our investors, we are making a real difference for frontline workers and addressing the critical labor shortage in manufacturing today.”
So what is a “physics-based” robotic system? GrayMatter contrasts its approach with the purely data-driven methods employed by others. The company explains:
Consider the problem of predicting a process output based on an input. If the output is expected to increase with increasing input, then the underlying model space is limited and can be trained with a smaller amount of data. There is no need to consider arbitrarily complex models. On the other hand, more complex representations and associated solution generation methods for handling constraints are required to produce acceptable computational performance. A simple neural network cannot be trained using the observed input and output data. In this case, there is no guarantee that the process constraints hold if the outputs used during training are noisy.
Interest in the company has fueled growth: GrayMatter appears regularly in our robotics job postings, and our May roundup had 20 jobs listed, the most of any job posting.
That growth is being fueled by continued fundraising. On Thursday, GrayMatter announced a $45 million Series B round led by Wellington Management with participation from NGP Capital, Euclidean Capital, Advance Venture Partners, SQN Venture Partners, 3M Ventures, B Capital, Bow Capital, Calibrate Ventures, OCA Ventures and Swift Ventures. The round is nearly double the $25 million Series A the company closed in 2022.