The record shows a turtle. In February 2026, roughly 12 hours after Elon Musk announced that SpaceX would pivot toward the Moon rather than Mars, Jeff Bezos posted a black-and-white photograph of a turtle stepping out of the shadows on X, the social network Musk owns, with no text at all. As Ars Technica’s senior space editor Eric Berger read it, the animal was the Blue Origin emblem, a nod to the fable in which the patient tortoise beats the faster hare, and a reminder of Bezos’ decade-old “Welcome to the Club” jab. The message underneath: slow and steady, and Blue Origin believes it will win the race back to the Moon.
Two landers, one deadline
The rivalry now has a formal scoreboard. After Artemis II flew astronauts past the Moon in April 2026, the first crewed lunar flyby in 50 years, both companies hold multibillion-dollar NASA contracts to land people there by the end of the decade, the first arrivals since 1972. SpaceX is building a Human Landing System based on Starship: 165 feet tall, the equivalent of 15 stories, with an elevator to carry crew and cargo down from a cabin near the top. Blue Origin’s Blue Moon looks more conventional but would carry lidar sensors to map the surface and pick the flattest landing spot. Both landers are meant to be reusable, to cut the cost of repeat missions.
The calendar is the real referee. Artemis III, scheduled for mid-2027, may test both landers’ ability to dock with NASA’s Orion capsule in orbit, a rehearsal for the crewed landings planned for 2028. NASA has also sketched a permanent lunar base with crewed landings every six months, an effort it says “will strengthen American leadership in space, usher in scientific discoveries, and serve as the proving ground for crewed Mars missions”. The competitor setting the clock is China, which plans its own crewed landing by 2030.
The scoreboard so far
On raw output there is no contest yet. SpaceX, started in 2002 with Blue Origin two years its senior, has flown Falcon 9 since 2010, ran 165 launches last year, 85% of all American orbital flights, and operates some 10,000 Starlink satellites, the largest fleet ever. Blue Origin reached orbit with New Glenn in 2025, lost that first booster, then landed one in November after dispatching two Mars probes on NASA’s Escapade mission; Berger counts two orbital flights against more than 600 for SpaceX. The encouraging part, for Bezos, is a new plan. Documents obtained by Ars describe an accelerated architecture that skips orbital refueling: an uncrewed demo using three New Glenn launches, two transfer stages and a Blue Moon MK2-IL lander bound for an orbit around the Moon of 15×100 km, then a crewed version needing four launches. Open questions remain, including what the transfer stages actually are, and the uncrewed MK1 mission must fly first, likely by late spring or summer.
The AI wildcard
The quieter prize is computing. Both firms are studying data centers in orbit, where solar power is cleaner and earthly protests and permits do not apply, though launch costs make the economics brutal. The paperwork is already moving: SpaceX has asked the Federal Communications Commission for clearance to fly as many as 1 million AI-computing satellites, and Blue Origin has requested nearly 52,000. “Orbital data centers are the most efficient way to meet the accelerating demand for AI computing power”, SpaceX wrote in its filing. Bezos, speaking at a tech conference in Turin, put it more broadly: “Space will end up being one of the places that keeps making earth better. It already has happened with weather satellites. It’s already happened with communication satellites”, he said. “The next step is going to be data centers and other kinds of manufacturing.”
He has been on this line for years. Unveiling his lunar plans in 2019, he said “It’s time to go back to the Moon”, and the turtle says he still means it.
Sources
- The Bezos-Musk space rivalry is shooting for the moon, and the winner could shift AI infrastructure | Fortune
- Why is Bezos trolling Musk on X with turtle pics? Because he has a new Moon plan. – Ars Technica
