I’m not sure I understand when you say “units” of power. I have almost 6,000sqft of conditioned space with two heat pumps. My power bill averages around $120 a month. How does that compare with what you guys are seeing ?
They mean kWh, I think, but some are so confused by misreading your original post it's hard to tell what they're saying.
I'd start small. That way you'll make $800 mistakes instead of $8000 ones. The first solar kWhs you produce each day are the cheapest, ones you store and use at night cost much, much more.
I'd reconsider net metering. It's usually a sweetheart deal for early adopters. And you'll probably get grandfathered in when more people sign up and they're forced to change to more realistic terms.
If you're dead-set against net metering, I'd recommend buying only enough panels for daytime loads at first. Let the grid fill in when clouds pass over, and at night. You'll over-generate at times, but it's OK to waste a little. Your mom won't scold you about power-starved children in Africa.
In time you might figure out ways to soak up that excess generation. If you have a well maybe pump water to a holding tank on a high spot. Maybe run the A/C before noon to pre-cool the house so you won't have to run it as much later when the sun gets low and panel output drops. Maybe get an EV on a smart charger programmed to only use excess generation. Lots of possibilities. Maybe even add a small battery to experiment, after you've grabbed the low hanging fruit.
If you want to kick self-consumption up to 90% or so you'll have to do more analysis and spend more money. Let's say 50 kWh/day average is your seasonal peak. And let's say 12 kW of panels at your location can supply 50 kWh/day on average in August. That might cost $8-10k (plus your free labor, ha). Now let's say 20 kWh of your daily consumption is after sunset. That's 25 kWh of LiFePO4 battery (80% DoD), so add 10k or so. Plus power electronics, cabling, etc.
These are just guesses, but ballpark. You need to study your actual consumption patterns and use
the PV Watts calculator to estimate monthly panel output. That gives you a starting point, but does not address daily variation. So you'll still have days when you throw power away and others where you need the grid.
Design the system so you can easily expand it after you've spent a year or so gathering data. But in general going after that last 10% is very costly. At least you already have a generator, so you don't need a 100% solution.