I agree with both of you - mostly.
Nizam's right - over the last few years we've encouraged several existing EA customers to move over to a cloud implementation, and with the help of Nizam and his excellent team, this is going well. Nizam's approach of a hands-on, thoughtful trial of EA-in-the-cloud works well. (Most of my customers look at it for 5 minutes, then decide "it's just like my normal EA - only faster!")
But what about the new EA users?
More and more of them are in this 'cloud only' world, so they will never be able to experience EA because they can't get it installed on their desktop. I recently waited 5 months for a UK government customer to get EA installed on one of their machines - they also have a 'cloud only' policy, but thankfully some EA evangelists to push for the change.
I'm concerned that EA might get left behind for these new customers: one has already asked my if Sparx are serious about SaaS, given that there isn't a way to do an instant trial.
If we're serious about EA SAAS, then lets go the whole way, and offer a 30-day, or 10 day, or 5 day free trial of EA SaaS.
Takeshi is quite right - someone will have to pay for the hosting, and the logical funder would be Sparx HQ.
Partners might be happy to fund it, so long as they get a margin on any EA licences and hosting revenue, but of course the customer might go to Nizam direct, or host it in their own cloud and buy direct.
So it's back to being HQ-funded.
I can't see the cost of this being huge, compared to overall EA revenues (several $10M s) : more of an issue would be to fund and manage the setup and maintenance of the trial environment, and designing it to have some level of security, but minimal cost.
It's going to be hard to sell EA SaaS to new customers without this offering.