I wonder how far away this is ……..2024 ?
Forgive my Slick Willy paraphrase, but it depends on what one's definition of "predictable" is. I'd contest that it largely is already. The infrastructure is pretty mature (obviously they're still building out V2M, but they can also do basic math on burn rates) and the recurring cost to maintain infra (launching more sats to replace dying ones, etc.) is pretty straightforward. The cost to buildout gateways is understood. The unpredictable ramp on users is in the past; the future take rate should be reasonably calculable.
The fundamental issue--and you can go back years in this forum and find I've made countless similar posts so sorry about beating the horse here--is that the practical market opportunity for Starlink has always been a) those who do not have reasonable terrestrial service AND b) those who can afford to pay for connectivity service. Look around the world and there's not
actually a ton of material revenue generating opportunity there beyond folks who already pay the stalwart space industry for shitty satellite internet. So, right now, Starlink is more or less resigned to playing at being a big fish in a small pond. They're certainly doing an excellent job at it (mind, fish in a barrel isn't exactly difficult; of COURSE Carnival wants to provide better-than-dialup service to their customers), but the cold reality is that there's really
not a ton of growth opportunity to expand well beyond what SX is doing now.
If you're looking for a step function in the above, which IMO is really what Elon is talking about with "predictable", it is once there's a critical mass of V2 sats on orbit. To get there we need a properly operational SS and a coupled thousand V2's distributed fairly symmetrically. (Looking at SS development pace, I'd probably pin this toward the end of the decade--maybe 2028-2029?). The big enabler here is that the bigger V2 satellites will have bigger antennas and, because Physics Is, those bigger antennas can form smaller beams. Smaller beams means SX will be able to provide significantly more capacity to a given area on the ground and THAT means they can provide [acceptable] service to a larger number of customers. Assuming SS+V2 can enable a lower per-capacity-unit cost than the current F9+V2M burn rate (which they will ultimately achieve, though its not clear exactly when), this will allow SX to become more price competitive with the terrestrial services and thus start to convert some of the comcast/spectrum/etc customers out there that are just happy to spend ten bucks less a month.
The big caveat here is that, in general, "better" starlink service than what's currently provided (more capacity and/or lower latency) ISN'T much of a needle mover when it comes to appealing to the majority of terrestrial service subscribers. Sure there are folks that are happy to pay more for better service, but for the lion's share of revenue generating customers, It really all boils down to the bottom line of "are you cheaper than the other service?". THAT's the major growth opportunity for Starlink.