SpaceX is poised to fire off a fresh Falcon 9 rocket on Thursday, delivering a comically tiny payload for Taiwan’s National Space Organization. At 1,047 pounds, the Formosat-5 Earth-observing satellite is almost light enough for a human to deadlift—but it’ll launch atop a Falcon 9 rocket with 50 times more capacity. The overkill is thanks to a years-long delay, and SpaceX will take a substantial financial hit to make good on a contract it signed in 2010.

Elon Musk’s space flight company will attempt to launch the rocket from Vandenberg Air Force base in California during a 42-minute window opening at 11:51 am PDT on Thursday. The satellite is bound for heliosynchronous orbit, where it will pass over Taiwan every two days for data retrieval. After the lightest single payload to ever hitch a ride on a Falcon 9 separates, the booster will fly back for a drone ship landing—hopefully to be reused in future, more economically viable missions.

The Formosat-5 is Taiwan’s first satellite designed and built entirely with the nation’s resources. More than 50 teams from across the country built it to facilitate academic research, disaster prevention, and humanitarian assistance. Originally, the sat was supposed to fly on SpaceX’s Falcon 1e, an upgraded version of its first orbital-class Falcon 1 rocket with a lift capability of 2,200 pounds. And according to industry analysis site Space Intel Report, they paid $23 million for the privilege—compared to the typical $62 million for a commercial Falcon 9 launch today.

So how did Taiwan hitch a discounted ride on a Falcon 9? Delay after delay. SpaceX hoped to deliver Formosat-5 as early as 2013 by launching the small-lift vehicle from a pad on Omelek Island in the military-controlled Kwajalein Atoll southwest of Hawaii. But because of low demand, the company scrapped the single-engine Falcon 1e in the summer of 2011, refocusing its efforts on developing new versions of the Falcon 9. Regardless of how SpaceX launches Formosat-5, the price remains firm for the customer.

After the Formosat-5 mission was moved to the Falcon 9, SpaceX targeted its launch for 2015, searching for another paying customer to share space onboard. Launch middleman Spaceflight was in talks with SpaceX to fly its Sherpa rideshare spacecraft on the mission. The Sherpa is a tow vehicle and deployment system for smaller, privately owned satellites; Spaceflight would have flown 90 of them, many from Earth-imaging company Planet, on the Formosat-5 mission.

But the mission suffered even more setbacks after SpaceX’s Falcon 9 rocket and Dragon spacecraft was destroyed during a launch to the space station in June of 2015. Then, the following summer, SpaceX lost another Falcon 9 after an explosion at its Cape Canaveral launch pad. The incident forced Planet to pull its satellites from Sherpa and launch on an Indian rocket instead—leaving Taiwan’s National Space Organization as the sole passenger on Thursday’s mission.

Despite the delays, Taiwan didn’t opt for another rocket provider like Orbital ATK, which operates the Minotaur rocket for missions to low-Earth orbit at a cost of around $30 million. Instead, SpaceX will pay 1.25 percent of the launch costs back to them for every month that Formosat-5 is delayed, according to the mission’s contract.

So how much is SpaceX going to lose on this mission? If you remove the potential reusability of the Falcon 9 booster for a moment, a lot. According to a launch cost analysis by investment firm Jefferies International, SpaceX usually makes a 40 percent profit from $62 million commercial Falcon 9 launches with new boosters. That puts $25 million in the bank and $37 million toward direct launch costs. With Taiwan’s severely reduced fare of $23 million, SpaceX is not only foregoing its profit but will be out-of-pocket for the remaining $14 million.

Thursday’s launch from Vandenberg could have provided an opportunity for SpaceX to take another step on its path to reusability. Given the light payload and short flight of Formosat-5, the Falcon 9 booster would have more than enough fuel to navigate back to land for the first ground landing outside of Cape Canaveral. But SpaceX doesn’t have the proper approval to use its California landing pad just yet. Instead, the booster will come flying back to the Just Read the Instructions robotic ship parked in the Pacific Ocean.

More on SpaceX

Space

SpaceX is rebooting its colonization plan, and may pivot to focus on a moon base that would aid that effort.

Space

What does this mean for SpaceX? What does it mean for the future of commercial space flight? What does it mean for Internet access in Africa?

Space

SpaceX plans to launch nearly 12,000 satellites to create fast internet access around the globe. But is that all it will do with its constellation?

The loss-leading mission puts all the more pressure on SpaceX’s strategy of reusability. SpaceX indicated it will also attempt to recover the rocket’s payload fairing—which encapsulates the satellite—during takeoff, which is worth around $6 million. Elon Musk says that a factory fresh Falcon 9 booster accounts for 70 percent of the $37 million in direct launch costs which totals to around $26 million. If SpaceX can reuse that booster enough times to pay for itself, the long-term loss from Formosat-5 will not be as significant.

For an industry struggling to reduce the price and accessibility of space, the Formosat-5 launch is a dramatic waste of resources. It also reveals an emerging but underserved market for customers hoping to launch smaller satellites. Vector Space Systems—founded by former SpaceX exec Jim Cantrell—is promising a dramatic reduction in launch costs for smaller payloads with readily available flights. It completed a test flight of its rocket with commercial payloads earlier this month. And launch startup Rocket Lab, with $75 million in funding, recently tested a rocket at its facility in New Zealand.

SpaceX wasn't built to serve the small satellite sector. But if it can perfect a rapidly reusable Falcon 9, it could manage to schedule multiple small-sat payloads on a single mission. After all, Elon Musk is certainly going to want a piece of what is expected to be a $7.5 billion industry.