I understand that (after the early stage), we will assemble gateway and have starship flights to the moon.
That is not certain. SpaceX won a contract to develop a Human Landing System based on Starship and perform two demonstration missions.
After that, a new contract will be awarded for operational flights. This selection process will be open to all bidders, just like the current one was. It is entirely possible that the National Team or Dynetics will apply again with a modified design, or even someone completely new out of left field. It is completely possible that this contract will be awarded to somebody else than SpaceX, or to two bidders.
It is actually very likely that NASA will want to award at least two contracts for dissimilar redundancy, the same way they do with Commercial Resupply Services (SpaceX Cargo Dragon, Northrup Grumman Space Systems Cygnus, and Sierra Nevada Corporation Dream Chaser) and Commercial Crew (SpaceX Crew Dragon and Boeing CST-100 Starliner).
In fact, NASA wanted to select two applicants, but because Congress allocated far less funds for the HLS program than NASA needed, they actually could not afford any of the bids. SpaceX was by far the lowest bidder, so NASA approached them about modifying their bid to move some milestones and payments around, so that they barely fit into the budget.
But I am not sure if we know what will we be carrying to the moon?
NASA asked for a Human Landing System that can transport 2–4 astronauts from the Lunar Gateway or Orion to the surface of the Moon, stay there for a specific amount of time (a couple of days to a couple of weeks), and bring them back to the Gateway or Orion.
That is all NASA asked for, and therefore, that is all NASA planned for.
The fact that SpaceX offered a vehicle that does far more than what NASA asked for, is very interesting, and it was also called out explicitly in the Source Selection Statement, but if NASA truly wants to keep the bidding process for the next stages open and fair, they cannot all of a sudden modify their mission plans to tailor them exclusively to Starship. That would also shut down any options for dissimilar redundancy in the future.
A 100 ton payload, of what? How many refuels?
There will be no 100 ton payload. The Artemis mission architecture has the HLS going between the Lunar Gateway and the Moon and back. There is no way of storing 100 tons of cargo on the Lunar Gateway.
As I mentioned above, the additional capabilities that Starship as the HLS brings to the table over what was actually asked for, were explicitly mentioned in the Source Selection Statement as an advantage. However, the Statement also points out that it might not be possible to make use of these capabilities due to limitations in other parts of the Artemis architecture, such as the capacity of Gateway.
It is noteworthy that SpaceX has already been awarded a Gateway Logistics Services contract to supply the Gateway using Dragon XL. Since Starship is going to return to LEO for refueling anyway, it could just pick up the cargo there, and SpaceX wouldn't need Dragon XL at all. (That's not quite true, since Dragon XL is planned to stay docked to Gateway for extended periods of time to extend the capacity of the station, whereas Starship would of course undock from the Gateway and land on the Moon.)
Even more crazy: Starship is much bigger than Gateway anyway, so why even build Gateway in the first place?
The number of refuelings is not decided yet, as far as I know. NASA would like to minimize the risk and keep the number of refuelings to 1, ideally. What NASA is proposing is that multiple tankers are launched, and that one of those tankers gets refueled by the other ones, and this tanker then refuels the HLS.
That way, the number of risky refueling operations between HLS and the tankers are minimized, and if something goes wrong while refueling the tanker, at least the HLS is not in danger.
Are we going to have robots on moon first?
Yes. The first six contracts to send landers and robots to the Moon have already been awarded to Astrobotic (2 contracts), Intuitive Machines (2 contracts), Masten, and Firefly under the Commercial Lunar Payload Services (CLPS, pronounced "clips") program.
Some of the payloads that are going to be delivered by these providers are MoonRanger (a rover), VIPER-1 (a resource-prospecting rover), and PRIME-1 (an ice-harvesting drill). All in all, 24 payloads have already been announced by NASA.
What kind of habitat are we going to build? What will the first astronauts do there and in gateway?
NASA has not yet detailed any plans for the surface operations beyond Phase 1. Phase 1 consists of a 7-day stay on the surface with 7 EVAs, and then a return to the Gateway or Orion.
NASA has actually put itself into a pretty awkward position with this selection, but given that SpaceX was the only bidder that even remotely fit within the budget constraints set by Congress, they had no other choice. Once Super Heavy and Starship are fully crew-rated for all phases of flight (including liftoff on top of Super Heavy, atmospheric reentry from interplanetary velocities, including the belly flop and the flip'n'burn, and in-flight refueling), almost every single part of the Artemis mission architecture is obsolete: there is no need for SLS, no need for Orion, no need for Gateway, no need for Dragon XL. Also, no need for the ISS (a single Starship has the same habitable volume).
So, by Congress reducing funding for the HLS, and thus forcing NASA to fund Starship, NASA will now be funding the very thing that will make Congress' favorite rocket, the SLS, as well as many traditional launch providers obsolete – not just in the US but in many of the Artemis partner countries as well. (For example, Ariane will be obsolete, as will be the two Gateway modules provided by ESA and ESA/JAXA.)
Note that one major advantage of SLS, and one major disadvantage of SpaceX from NASA's point of view is that SLS, through its vast network of contractors, sub-contractors, sub-sub-contractors, sub-sub-sub-contractors, partners, partners-of-partners, and so on, provides jobs and pays taxes in almost every State of the US. And the more money they waste, the more jobs they generate, and the more taxes they pay. SpaceX is exactly the opposite: they are lean, mean, and efficient. Hence why Congress provides far more funding for SLS than they do for HLS.
NASA, being a government agency, is dependent on support from Congress.