We are using the HMC465LP5 in our application; the bias to the device is provided by the HMC981 as per the recommended application circuit in the HMC981 datasheet.
We have noted that when the HMC465 bias is OFF (e.g. EN=0 on the HMC981), the gate voltage VGG1 on the HMC465 is ~ -2.5V; this is Vneg on the HMC981.
However, the datasheet for the HMC465LP5 states that the maximum range for VGG1 is -2V to 0V (see page 8-4 of the HMC465 datasheet, Absolute Maximum Ratings).
Should we be concerned that the -2.5V is applied to VGG1? The two datasheets seem to contradict each other.
Regarding your described usage of the HMC460LC5, HMC463LP5, and HMC465LP5, I discussed this situation with one of our designers (KB) who is familiar with this type of cascode distributed amplifier architecture. He confirmed that the parameter of concern would in this case be the gate to drain breakdown voltage. For a given Vdd the gate to drain voltage would be greater when using Vgg1=-2.5V than when using Vgg1=-2.0V. For that reason -2.5V would be more stressful than -2.0V. However, the Vdd-into-HMC981 values you're using would result in the devices' nominal Vdd voltages being applied at their Vdd pads (i.e. 8V, 5V, and 8V, respectively) such that the drain of each amp would be biased approximately 1V below their Vdd AMR values. Vgg1=-2.5V is only 0.5V below its AMR and that 0.5V does not consume the entire 1V of margin present for Vdd. There is likely additional margin, but to determine exactly how much would require a more detailed analysis.
In this case we expect that you can safely proceed with using Vgg1=-2.5V without suffering any short term or long term consequences.