The simple answer is: Nobody knows. The more complicated answer is: Lots of experts have made projections of how widely the epidemic could spread, but the number of uncertain factors involved (such as how well the medical infrastructure holds up, how many beds are available, how quickly new cases are reported and isolated, and how safely bodies are disposed of) mean that the range between the best and worst-case scenarios is vast.
Two sets of projections were made in late September. The WHO, in the same New England Journal of Medicine paper, projected that if the rate of increase didn't change, there could be 20,000 cases by early November. That may or may not be accurate: it's more than the current reported number of cases, but could accurately reflect the situation if (as the WHO suspects) the reporting infrastructure has collapsed in several of the worst affected countries.
At around the same time, the US Centers for Disease Control and Prevention (CDC) made a longer-term series of projections for the countries of Liberia and Sierra Leone that goes up until 20 January 2015. These took into account how successful the efforts to contain the disease could be. If the efforts were largely successful, the CDC predicted, the epidemic would be virtually over by that date. If the efforts were entirely unsuccessful or non-existent, then there could be as many as 1.4 million cases by then.
The point is not to say that one or the other projections may be right – they're not predictions as such, just models to guide policymakers towards the best decisions. Rather, they serve to point out the incredible amount of uncertainty when dealing with rapid outbreaks of infectious diseases.