🚨 Does Google even have enough compute to release Gemini 3.5 Pro?

• The model has been delayed for months, as it apparently "failed to meet Google’s internal targets", but is the actual reason that they aren't able to sustain due to compute constraints?

• It's been 5 months since 3.1 Pro dropped and was the best model at the time, how have they not been able to drop a model since that is good, especially when Anthropic & OpenAI have been dropping models every ~6 weeks that are improvements upon the last and that is with the US intervention too?

Do you think it is purely incompetence, or that they have a model ready but can't supply it comfortably so are holding off?