"Algae fuel yields have not yet been accurately determined, but DOE is reported as saying that algae yield 30 times more energy per acre than land crops such as soybeans [Soy: 59.2-98.6 gpa in Indiana], and some estimate even higher yields up to 15000 gpa ." sourceI think the first link HoosierRon provided is the most interesting.
They claim 5-10 times the yield of other crops for biomass. Assuming they mean ethanol, at worst case yield that's corn yield 354 gallons per acre times 5 = 1770. The best case I found online was for sugar beets, at 714 gallons per acre. Multiply that by ten, and you have 7140.
$25,000 installation cost for a 1770 gallon annual yield is a hard sell, at least to me. You aren't getting 1770 gallons of ethanol from that, you're getting biomass that can be converted, at some additional cost, to ethanol. Even if you get the equivalent of $1 per gallon from your yield, it takes over 14 years to pay for itself plus your annual costs in maintaining it, harvesting, supplying water, and so forth.
On the other hand, 7140 gallons per acre per year changes things a lot. Assuming the farmer gets $1 per gallon equivalent, that pays for itself in less than 5 years. After the system is paid off, it would probably be, by far, the most profitable crop in the world outside of illegal drugs.
One square mile farmed that way would be 640 acres * 7140 = 4.57 million gallons of fuel per year. The US currently uses about 146 billion gallons of gasoline per year. Ethanol has about 70% the energy density of gasoline, so to offset 146 billion gallons of gasoline we would need to use about 211 billion gallons of ethanol. That would require just over 46,000 square miles to offset. That's $740 billion to install, and a total area of 1.5% the surface of the continental US. That's enormous, but not impossible... and the amount of money required is huge, but likewise not impossible.
Assuming they can get somewhere near their optimistic projections, they are sitting on US energy independence.
So it looks like your 1,770 estimate would be the rock bottom possibility.