Roughly 3 acres per megawatt done right. (Allowing paths between arrays for maintenance, etc.) A lot of solar farms are 5-10 acres per megawatt due to idiots wanting to prove that $4+/W thin film modules are somehow better than <$3/W (turnkey installation price) polycrystalline modules.
I feel very strongly negative about:
1. Pricing per MW nameplate/peak vs LCOE amortized cost including availability, OM, and replacement when comparing sources.
2. Determining area based on peak/nameplate power, rather than area per yearly average MWhr
3. -not- including storage costs (which underprices the real cost of solar and other intermittent sources as it basically puts that cost on someone else)
4. Comparing / calculating break even costs based on the current end user price per kWhr -rather- than the feed-in tariff paid back for surplus power (see 3).
When the above are taken into account, solar sucks...in all ways...and will always, even if it achieved the 40-50+% maximum quantum efficiency when compared apples to nuclear.
See. I -like- the environment, and if there is a way to provide all the energy we need, and more, at low cost, forever, while covering as little of the planet as possible...I'm for that.
The -only- way solar will ever be truly competitive with nuclear AT CURRENT COSTS is if launch costs drop below $300-500/kg or we can mine materials from the moon autonomously, thereby enabling SPS (I did a recent study and we are within a factor of <5-10 on this -now-, which bodes well). That said, if you took the reigns of over regulation and other costs off nuclear, it's doubtful even at zero launch costs SPS would compete.