Todd Thomas, co-founder of Woodchuck, joins Sunny Ray to explain how artificial intelligence is turning construction waste into renewable energy. Woodchuck mounts computer vision cameras on job site dumpsters to identify and separate wood, cardboard, plastic, and metal in real time, diverting materials that would otherwise be buried in landfills. The company grew out of a partnership with Northstar Clean Energy, a Michigan utility that needed a steady new source of biomass to move away from fossil fuels. Thomas describes Woodchuck's first project at an Amazon distribution center, where the team diverted 658 tons of wood in one month, and how an early snowstorm exposed flaws in the AI that led to building a dedicated outdoor innovation lab in Grand Rapids, Michigan. He explains how sorting materials at the source, rather than after they are mixed, cuts hauling costs by 30 to 40 percent while creating new revenue for contractors. Thomas also discusses Woodchuck's recent seed round, its ongoing Fundify raise, and his upcoming book, Starving for Innovation.
An AI-powered smart dumpster spots wood in construction waste and turns it into renewable energy instead of landfill.
What is Woodchuck and what problem is it solving?
We're addressing two huge global problems at once. We have too much waste filling up landfills, and at the same time an exploding need for more energy. We work with construction and manufacturing companies and use AI to identify wood and separate it from the waste stream, then process it into biomass to generate renewable energy. We're reducing waste and increasing energy supply simultaneously.
Can you walk me through how a customer problem with Northstar Clean Energy turned into a company?
Northstar was our initial client and investor. They're the clean energy arm of a Michigan public utility with bioenergy facilities that needed more biomass and wanted off fossil fuels entirely. They needed a net new source since forestry and agriculture were already tapped. We realized 41 million tons of wood go to landfills every year in the US, so we built a way to capture that supply instead.
What did the first pile of construction waste you inspected teach you that a spreadsheet couldn't?
Our first project was an Amazon distribution center north of Detroit. In the first month we diverted 658 tons of wood and sent it to Northstar for electricity. We sent Amazon a sustainability report, and they asked about cardboard, plastic, and metal too. That pushed us to teach our algorithms to identify all those materials, and over three years our system got very accurate.
What was the hardest early assumption about the wood waste stream that turned out to be wrong?
We built our MVP in about 90 days and rolled it out on a live site. The next morning it snowed overnight and our data was a mishmash because the AI had no idea what to do with snow-covered material. We built an outdoor innovation lab in Grand Rapids with ten dumpsters to capture images in every weather and lighting condition and retrain the AI on real world conditions.
How does the AI actually see a waste stream and handle contamination that could ruin a biomass load?
It's really negative sorting. If a camera is on a wood dumpster, we teach it to flag anything that isn't wood and notify the right people to pull it out. We can even detect contaminants like fire retardant or preservatives that humans can't see, which matters because treated wood can't be used for energy, though we can still divert it for other industrial uses.
You frame sustainability as a cost cutter rather than a cost. Where does the 30 to 40 percent hauling savings actually come from?
Two areas. First, we consolidate material on site and move it by full size semis instead of small dumpster hauls, which dramatically cuts transportation cost. Second, instead of paying tipping fees to bury material, we sell it and share that revenue back with the contractor. Combined, that typically produces a 30 to 40 percent cost reduction for the job.
You raised 3.75 million in seed funding in mid 2025. What did that round buy, and what does the next one need to prove?
It let us build our permanent biomass processing facility and full-scale innovation center in Grand Rapids, so people could see the AI working at scale, processing 100 tons of biomass an hour. We just went live on Fundify for our next raise. Now we need to prove we can operate across many data center projects at once, not just one.
If a midsize general contractor wanted to start next Monday, what's step one?
Find us at woodchuck.ai and reach out, or find me, Todd Thomas, on LinkedIn. If you have a project starting soon, give us a call and we'll do a waste audit, look at your volume and material mix, and give you an estimate of how much money we think we can save you.
Building something daring? Sunny talks to founders like this every day. Fifteen minutes to see if your story belongs on the stage.
Claim your pre-interview