In large part because it's night half of the time, solar has a pretty low capacity factor, like in the 20% range (see https://en.wikipedia.org/wiki/Capacity_factor for details) so it matters whether this is 800MW peak or 800MW average. I think it's peak, which means that we're talking about something in the range of US$400M to US$500M worth of solar panels at current prices — which are the same prices as in 2014, when bids were twice as high! With a 35% capacity factor (Dubai is equatorial and sunny, although even Arizona only gets 19%), this project will have only 800 MW * 35% * $ 0.0299/kWh = US$73.4M/year of revenue, which is only a 16% return on investment — or ⅓ that, at 5⅓%, if we assume that the non-module costs of the project amount to ⅔ of the total cost, as shown in the bargraph on the article.
So, is the winning bidder (strangely, not named in the article!) really willing to accept a 5⅓% IRR for this investment, because PV solar park competition is really that cutthroat? Or are they betting that by the time they have to buy the modules, the prices will have fallen (returning to their former exponential decline?), and/or that the notorious "soft costs" that eat up half the project cost are becoming cheaper?
There's no real floor on the cost of silicon photovoltaic modules, since almost all of the cost of producing them from raw materials is energy, not the materials themselves.