I'm the founder of AgFunder. We've worked with a bunch of agtech/hard tech companies and it's been very very frustrating taking them to VCs because they don't fit in their box. We had one company called SWIIM which we called the AirBnb for water that was creating a market for water and in doing so helping to fix a major issue of properly valuing water--enormous market, great unit economics, but there was hardware and deployment and it didn't look anything like VCs normally invest in. Another company we have on now called Autonomous Tractor Company (ATC) which is basically Cruise Automatic for tractors... There are 500,000 used tractors out that they can turn into true driverless tractors (not just GPS assist). VCs can't get their heads around the fact that they don't actually have to build tractors and that this is actually very capital efficient. Kudos to YC which gives great signalling to many misunderstood companies.
A lot of this stuff takes the form of government grants and university collaborations. Check out CMU's National Robotics Engineering Center. If I remember right, they were at one time partnering with John Deere to build autonomous tractors.
