Last year, SpaceX founder Elon Musk shared plans for his transportation system to send humans to Mars in the 2020s. But the fantastically huge rocket, with 42 Raptor engines and enormous technical challenges, seemed more like science fiction than reality. Then there was the small matter of who would pay the tens of billions of dollars to develop a rocket that had few—if any—commercial prospects beyond sending 100 people to Mars at a time.

Musk seems to have realized that his ambitions were a tad too ambitious in recent months, and has said he will release a “revised” plan for Mars colonization that addresses some of these technical and fiscal questions. Now, we know this discussion will come during the 2017 International Astronautical Conference in Adelaide, Australia, on September 29. And this weekend, Musk dropped a big hint about the change.

In response to a question on Twitter, Musk wrote, “A 9m diameter vehicle fits in our existing factories …” And this is actually quite a substantial hint, because the original “Interplanetary Transport System” had a massive 12-meter diameter. By scaling back to 9 meters, this suggests that Musk plans to remove the outer ring of 21 Raptor engines, leaving a vehicle with 21 engines instead of the original 42. While still complicated to manage during launch and flight, 21 engines seems more reasonable. Such a vehicle would also have about 50 percent less mass.

Why 9 meters matters

At 9 meters the revised Mars rocket would still be considerably larger than SpaceX’s current booster, the 3.7-meter Falcon 9 rocket. But it would be smaller than the most powerful rocket ever flown, the 10-meter Saturn V booster that launched the Apollo crews to the Moon.

Downscaling the Mars booster suggests that Musk may be bending toward reality. A 9-meter rocket means that it could be produced in SpaceX’s existing facilities, saving the company the expense of building a much larger factory. (Pragmatically, it could also be produced in NASA’s rocket factory in Michoud, La., without major renovations). A smaller, but still powerful rocket also opens the door to commercial opportunities and military contracts.

Most notably, the US Air Force is in the midst of soliciting bids for the second phase of a $2 billion competition to develop new launch vehicles that can meet the government’s space mission needs. This is part of the Air Force’s efforts to end US reliance on the Russian-made RD-180 engine, and this competition is for development contracts to build launch systems capable of flying missions by the early- to mid-2020s. It is possible, then, that SpaceX may bid for some of these funds to help develop the Mars rocket, perhaps for the Raptor engine, or the entire vehicle.

A successful Air Force bid would answer one important question Musk faces—how to pay for the Mars rocket. An answer to another key question could come later this year, whether SpaceX can really manage to control dozens of rocket engines during flight. Although the Falcon Heavy rocket has a different configuration from the Mars rocket, it requires the coordination of 27 Merlin engines during launch. If SpaceX can do that during the Falcon Heavy’s maiden launch—possibly later this year—then controlling 21 engines on the Mars rocket doesn’t seem to be that great of a stretch.