|
WK621: When AI eats your moat
January 2026 has been a long year. AI is eating software moats while enabling climate hardware companies to growing faster than we could have imagined 12 AI is eating software Here’s a conversation from our network that stuck with us this week. Someone asked why enterprise software multiples are compressing, and the answer reveals something profound about where we are. “Software moats are falling in a world where a potential competitor can quickly replicate what you have. Or in a world where customers churn more because they can build software themselves instead of buying.” The punchline: software may have been eating the world, but core competitive assumptions like switching costs, lock in, etc are getting vaporized. It seems like this will cause more founders and investors to pivot into places where more durable moats might be built. We think a lot of those will include hardware. This connects to a generational divide Aashay Sanghvi noted recently: some investors believe in luck and iteration and that you can’t know outcomes without taking lots of bets. Other investors increasingly believe success is engineered a priori. “You’re either in or you’re out, the setup at inception is everything. All hail the club deal.” We now have three approaches: “big swings” (call your shot), “fast swings” (call your customer), and “make kings” (call capital). The first two represent internally derived permissionless conviction. It’s very unclear that king-making is a thing that works most of the time versus a nice story for LPs. What This Means for Climate Good news, at least for hardware, at least for now. These are physical systems where the moat is atoms, not bits. While pure-software plays scramble to defend switching costs, companies building heat pumps, fire suppression systems, and electric bikes have moats that AI can’t code away. For now. Hardware moats come from places VCs have historically tried to avoid: energy constraints, component supply chains, regulatory limits, manufacturing complexity. These used to be called “barriers” for a reason. But in a world where software moats are vaporizing, those barriers start to look like defensible advantages. The hard stuff is becoming the durable stuff. That said, AI is also a gift for hardware teams. LLM platforms increasingly enable faster workflow automation, for more team members in areas ranging from logistics to marketing. So hardware teams can ramp revenue more quickly where the limiting factors are quickly becoming supply chains. The design iteration that used to take weeks now takes hours. Blessed be the hardware startups, indeed. Energy is a perfect example of constraints becoming moats. US electricity demand surged in 2025, and solar handled 61% of the growth. The cleanest way to handle AI’s power demands is already here, but only for those who’ve built the infrastructure. And speaking of AI’s footprint: most environmental impact discourse focuses on “median queries,” but what about a full Claude Code session? The benchmarks are illuminating. AI coding agents use real energy, which means the companies generating and managing that energy have something software can’t replicate. Hardware moats will be vulnerable in increasing ways soon too as AI is coming for design, component sourcing, integration, branding / packaging and more. We think the only moats that will remain are trust and a willingness to do what others are not or cannot – think boots on the ground and unique engineering. Portfolio Updates Sonic Fire Tech made waves (sorry) at CES 2026, earning recognition as one of the most innovative consumer products. Their infrasound technology that suppresses wildfires before they spread also landed a Scientific American feature. With California fire season becoming a year-round reality, this couldn’t be more timely. We also welcomed RTV to the portfolio. They’re working on recyclable stretch fabric with a material that’s historically been impossible to recycle. Fast Company has the story. Mill continues to expand their kitchen waste-to-chicken-feed model. Business Insider profiled Matt’s journey from iPod to Nest to Mill; the through-line being hardware that changes behavior at scale. Plentify secured their Series A to export their South African energy management solution internationally. Robot.com captures something we’ve been saying for years: “When people imagine robotics, they picture factories. The real opportunity is everywhere else.” Wild Technologies landed a TechNexus Connect interview about their electric trailer platform and how they’re reinventing towing, which turns out to be a surprisingly large slice of transport emissions. Opportunities Transit Tech Lab’s Partnership Fund is accepting applications through February 27 for early and growth-stage companies with solutions for MTA and Port Authority. Apply here. The Abundance House residency is Month-long program for impact-oriented founders. Applications now open at abundance.house. Hiring: Our portfolio companies are actively building their teams across engineering, operations, and finance. A few highlights: Mill is looking for a Lead Mechanical Engineer in San Bruno and a Senior Associate, FP&A. Nevoya needs a Dispatcher in San Bernardino and Regional Drivers in Phoenix. Circuit is hiring Driver Ambassadors in Washington, D.C. And Climatebase is looking for a Part-Time Graphic Designer. See all 223 open roles at jobs.thirdsphere.com. There is no time to waste. Let’s go 2026!
Best, P.S. WK621 = 621 weeks since we sent our first Third Sphere newsletter. P.P.S. Know founders we should meet? The companies thriving today were not working in popular areas when we invested. Our criteria are pretty simple – great team, bad logo, working demo, no customers and the potential to have a huge impact in some area of climate (probably the one that isn’t getting much media coverage right now). Did someone forward this email to you? Here’s how you make sure not to miss the next one. You can find our previous updates here. |
|
|