This value indicates the number of iterations required to obtain the sum of squared errors (SSE). The ARIMA algorithm performs up to 25 iterations to fit a model. If the solution does not converge, store the estimated coefficients on the Storage subdialog, and run the analysis again entering the column of stored coefficients in Starting values for coefficients. You can store the estimated parameters and use them as starting values for a subsequent fit as often as necessary.
The algorithm may also fail to converge because you include a constant in the model. You can try performing the analysis again without the constant.