At this scale a careful feasibility study by a specialist like a Moodle partner http://moodle.com/partners/ is a must.
Still a few points you should clarify either way:
Q1. Moodle and Mahara integrate well together. Still when you say SSO, do you have a certain technology in mind?
Q2. 100'000 potential users are not a problem, if they don't use the plattform intensively. Read the Performance documents mentioned in the introduction to this forum to see what heavy usage is.
Q3. Two years are a life time for hardware. That is plenty of time to grow together with the usage.
Q4. I find it is odd that for your boss the number of possible is _the_ criterium for two reasons:
1) What about the functionality? I mean, what are your users going to do resp. what does your institution want to give them? Have you done an evaluation? Have you compared Moodle 1.9 to 2.x? (See Tim's hint above.)
2) AFAIK Blackboard is software, you can buy any number of licences by throwing money. What about hosting? Both products need hardware, network infrastructure and have to be maintained. Moodle has a head start because there's no licence fee, you can invest that money on hardware and staff.
In addition:
- Do you have an idea about the amount of data the site will produce daily/monthly/yearly?
- How often you are going to take backups, where will they be stored offline?
- How much availability is required? How much down time is tolerated for which types of fail-over?
- How much network traffic do you expect, average over day/month, also the peak bandwidth. The expected/allowed latency from various service providers?
- User management?
- First level support for million(!) users?
This is a random list of things which came to my mind. Should work differently in a real case