Starting with Starship, flight 14 was a critical milestone. Starship made it to orbit for the first time and deployed operational Starlink V3 satellites. I think we have a recap video we're going to play.
For flight 15, which we expect to happen next month, we're going to attempt to catch both the booster and the ship. If we succeed, this will be the first time that an orbital launch vehicle has been fully recovered. This is a massive technology milestone. The only way that we can extend consciousness beyond Earth on a sustained basis is with a fully reusable orbital launch vehicle. So I think this is a profound technology milestone in the history of civilization consciousness. So let's look at what a simulation of flight 15 would look like if we succeed. If we don't succeed in flight 15, we'll get there. Success is not guaranteed on these first flights, but I'd say it's more likely than not that this succeeds next flight. And if not, we have a steady stream of Starship flight attempts through the end of this year and then accelerating next year. So with that, the flight 15...
We've also announced the development of the Louisiana spaceport. So that's going to be an incredible asset to the country and to the world. That will, I think, at some point probably have on the order of 20 or more launch towers and enable the transport of millions of tons per year to orbit and beyond. So that's really going to be a big deal.
We expect Starship to fly weekly at some point in the first half of next year, and then in the second half of next year twice weekly, or maybe even more frequently than that by the end of next year. The rate at which we launch Starship will increase dramatically as we solve rapid and complete reusability. So we'll have complete reusability. Then at first there will be a little bit of refurbishment that's needed between each flight, and we'll iron out the need for any refurbishment such that you can immediately land the rocket and launch it again, just like an airplane. As I said, this is a profound technology milestone that most people in the world don't realize how important it is, but it will be obvious in hindsight.
Starlink is also just getting started. Flight 14 deployed 26 terabits per second of Starlink bandwidth. That's equivalent to 10 Falcon 9 launches. So a single Starship launch, which wasn't even fully loaded — it was about maybe half loaded — it did the equivalent of 10 Falcon 9 launches. So what I expect over time, not in the distant future, probably within 3 years, is that we will have about 100 times the bandwidth of the current Starlink constellation.
So because we will have launched satellites with an order of magnitude more bandwidth, and we will place into orbit an order of magnitude more Starlink satellites — so probably something on the order of 100,000 satellites, each of which have more than 10 times the bandwidth of the current Starlink system. So I think revenue for Starlink will be at least somewhat proportionate to the total bandwidth delivered, and we will also have direct-to-cell, high-bandwidth direct-to-cell capability. And we will be building out our terrestrial network so that there's broad-based connectivity anywhere that SpaceX and Starlink are allowed to operate. So anywhere in the world that we're allowed to operate. And I think the demand for bandwidth is going to be dramatically higher than it is today. I think in 3 to 5 years there's probably a demand for a thousand times more bandwidth than there is currently, because AI or SI agents will, even on a per-agent basis, use far more bandwidth than a person, and there will be far more agents than there are people. So this ends up requiring a several order of magnitude increase in communication bandwidth in order to satisfy that communication demand.
So talking about SpaceX superintelligence, SpaceX SI, sexy, whatever. So as I think people are starting to realize, the constraint on deployment of superintelligence is more in power and cooling and infrastructure than it is in chips.
So what we've done at SpaceX is we've taken a team that has been able to achieve a market share in mass to orbit that is almost 90% of all mass to orbit of Earth, where China is about, I think, roughly 6%, and the rest of the world combined, including the rest of the US, is about 4%. So a team that has been able to achieve results an order of magnitude greater than all of our competitors in space combined, and applied that same team to the deployment of terrestrial compute. So this is a bit like the New York Yankees playing in little league. It's not fair, but the net result is being able to deploy power, cooling, infrastructure at a rate that is vastly faster than anyone else, and may at some point be faster than everyone else combined. So I'll just let that sink in for a minute.
So I've been spending personally a lot of my time in Memphis, often four days a week, night and day, just getting the Memphis supercomputer clusters online. We're tracking to be able to bring roughly 200,000 GB300s online per month for every month for the remainder of this year, and continuing that pace next year, with next year being primarily Vera Rubin compute. At this point, I would say I'm fairly confident of being able to bring an incremental 10 gigawatts of compute online in 2027. That would be in addition to the roughly 3 gigawatts, maybe 2.5 to 3 gigawatts that we'll have online by the end of this year. So I would put it this way: I would be surprised if we brought on less than 13 gigawatts of compute by the end of next year.
And as I mentioned earlier, power, cooling, electrical equipment, infrastructure is somewhat of a limiting factor. So to make sure that that is not a constraint for our buildout, we're actually going to aim for an incremental 20 gigawatts of power and cooling infrastructure next year, so that the amount of power and cooling that we have is far in excess of what is needed to turn the GPUs on. So there is a non-zero possibility — maybe it's a small possibility at this point, but it's not zero — of SpaceX having over 20 gigawatts of superintelligence compute online next year. A small possibility, but it used to be zero. It's not zero anymore.
We're also planning to have the most powerful training cluster online by the end of Q2, which will be 440,000 Vera Rubin computers deployed for optimal training. This will, I think, be extremely helpful for getting Grok to be potentially the best model, but certainly to be extremely competitive.
So on the Grokbot side, we're seeing the fastest growth of a product that I've ever seen in my life for Grokbot. I've never seen anything grow this fast. We're also seeing that the usage of Grokbot grows over time per user. It does not decline. So as we make Grokbot more useful, we're actually seeing users use it more per week, not use it less. And we will be deploying Grok at scale for enterprise customers, which is a very big deal. So it can be used effectively in teams by companies as though it were a virtual human. We already are in early deployment with some trusted customers, and we expect to be in wide deployment within a few weeks for enterprise customers for Grokbot.
I think Grokbot may end up being what uses most of our superintelligence compute.
So I mean, in a nutshell, what this implies for SpaceX annual recurring revenue by the end of '27 is somewhere between 500 billion and a trillion.
