Kepler’s decision to build its own cubesats surprises manufacturers

by and

Planet felt like it didn’t have a choice. When the Earth-observation startup began launching cameras into orbit in 2013 and 2014, manufacturing lines weren’t churning out cubesats by the dozens.

That’s no longer the case. Blue Canyon Technologies, AAC Clyde Space, GomSpace, NanoAvionics, Tyvak and several others are ready and willing to build cubesats en masse. So it came as a surprise to many cubesat manufacturers when Kepler Communications announced plans in January to manufacture its constellation of 140 Internet of Things satellites in-house.

Kepler is poised to become one of the world’s largest cubesat operators once its constellation is fully in orbit, a target set for the end of 2022. Only Planet currently operates a fleet that large.

Instead of formally soliciting bids from a wide range of cubesat builders, though, Toronto-based Kepler turned to the University of Toronto Institute for Aerospace Space Flight Laboratory (SFL) for help setting up its own manufacturing line. Kepler also received 1 million Canadian dollars ($760,000) from the Canadian Space Agency to mature its bus design and production techniques, leading some observers to conclude national pride could play a role. Through Kepler, Canada is establishing a robust cubesat manufacturing capability.

Exhibit A

As startups announced plans for cubesat constellations in recent years, companies expanded manufacturing lines to meet the demand. Denmark’s GomSpace scaled up manufacturing to fulfill a large order from Sky and Space Global that was suspended last year when the Australian Internet of Things startup struggled to raise enough money for a 200-cubesat constellation.

Blue Canyon Technologies is preparing to open a manufacturing plant this spring capable of producing multiple cubesats per week, according to CEO George Stafford.

Cubesat builders, such as Blue Canyon Technologies, say they are ready for bulk orders — if they ever materialize. Credit: Blue Canyon Technologies

Stafford and GomSpace CEO Niels Buus said both their companies have the capacity to build a Kepler-sized constellation. Even manufacturers that lack the means to build 140 or more satellites today could quickly scale up to churn out batches of identical satellites if given the chance, said Arnoldas Pečiukevičius, a systems engineer at NanoAvionics of Lithuania.

In spite of all that capacity, manufacturers aren’t getting the chance to bid on large orders of identical cubesats. Companies continue to order cubesats in ones and twos, said Craig Clark, AAC Clyde Space founder and chief strategy officer. If someone said to a cubesat builder, “can you build 100 satellites like the one you just did, they would bite their hand off to do it,” Clark said at the SmallSat Symposium in Mountain View, California, earlier this month. “But they’re not being given a chance to do it.”

Clark wasn’t talking specifically about the Kepler order but the market in general. Nevertheless, Kepler’s decision can be viewed as Exhibit A. AAC Clyde looked like the obvious choice for Kepler’s constellation because the firm, based in Glasgow, built three prototypes for Kepler’s. When it came time to bid on the full constellation, though, AAC Clyde couldn’t offer the right mix of “price, schedule and technology” for the constellation, said CEO Luis Gomes.

Manufacturers worry cubesat constellation operators have unrealistic price targets.

“Cost is only one variable,” Stafford said. “We encourage the idea that affordability, speed to market and on-orbit performance must all work together to achieve success.”

Due in part to the lack of volume orders, cubesat prices are closer to those of $400,000 performance cars than of $1,000 smartphones, Clark said. Even at those prices, manufacturer net profits tend to fall in the range of 5-10%, he said.

In the long run, some constellation operators may spend more to produce their own cubesats than to buy them from a dedicated manufacturer, said GomSpace’s Buus, since dedicated manufacturers can achieve higher volume production. “If you need 100 satellites a year, it will always be less expensive to build 1,000 satellites a year,” he added.

Pumpkin President Andrew Kalman said his company is not the right fit for cubesat buyers looking for the cheapest components.

“The industry is in a period of substantial influx of capital, looking for investment vehicles, and NewSpace companies are making convincing pitches to secure big money,” he said. “Because of the frenetic pace, I suspect objective assessment of quality, real lead times, design flexibility and design headroom, and a variety of other design, manufacturing and production metrics are sometimes lost in the excitement of dreaming about building the next big thing.”

NewSpace ethos

Kepler CEO Mina Mitry is confident, though, that vertical integration is the right approach for the firm’s Internet of Things constellation. Manufacturers aren’t ready to build Kepler’s constellation of 10-kilogram cubesats quickly, affordably and reliably enough, Mitry said.

Kepler evaluated around 10 potential manufacturers through a process that involved bids, on-site visits and technology reviews before deciding to build its cubesats in-house, Mitry said. “The evaluation criteria was price, performance and volume,” he said. “Finding that combination was quite challenging.”

Cubesat manufacturers who have invested in manufacturing facilities haven’t necessarily been handsomely rewarded, Mitry said. Without volume orders, they’ve stuck to their historical business models centering “around one-off contracting, subsystems manufacturing and that type of bespoke nature of development,” he added. “That hasn’t really allowed them to mature.”

Astrocast is building 100 satellites in-house (like the five shown here) to keep a closer check on cost, development cycles, testing, and other aspects of manufacturing, according to CEO Fabien Jordan. Credit: Astrocast

Mitry declined to name the manufacturers Kepler considered. GomSpace, Blue Canyon Technologies, Pumpkin and NanoAvionics said they were never contacted to bid on Kepler’s constellation.

SFL Director Robert Zee said Kepler’s approach is “typical of a NewSpace company,” which he defines as firms that provide services and manufacture their own satellites to keep costs low. SFL is designing and building the first cubesat in Kepler’s operational constellation and helping Kepler set up a facility to mass produce its remaining satellites. “The dimension that we bring is the experience, the heritage, the technology, [and] the high-performance satellites at low cost,” Zee said.

Kepler also is getting help from Canadian component supplier Sinclair Interplanetary, which granted Kepler a license to mass produce reaction wheels for its constellation.

To vertically integrate or not

Kepler’s case notwithstanding, some cubesat operators say they’ve struggled to find manufacturers to meet their needs.

Swiss startup Astrocast is building its 100-cubesat constellation in-house using 50 meters of lab space and five of its 40 employees, CEO Jordan Fabien said.

“Fully integrating our satellite production is the only solution we found to keep our development cycles as short as possible and make rapid and continuous design improvements,” he said.

In March, Astrocast will move into a new 350-square-meter facility to build as many as 30 satellites a year, he said.

Planet, with about 140 satellites in orbit, remains unconvinced dedicated manufacturers would serve it well.

“I’ve always said that if there are people who could build our satellites faster and cheaper than we could, we would absolutely prefer to buy those satellites from them and put our focus elsewhere,” Chester Gillmore, Planet vice president of spacecraft development and manufacturing, said in 2019 at an Aerospace Corp. event in Florida.

While Planet frequently updates its cubesats, it would prefer to outsource if a vendor could build its satellites without sacrificing Planet’s cost targets, service level or spacecraft quality.

“The industry would really have to advance to a high degree of maturity before that really becomes viable, but we would love that,” he said. “There is so much about what we’re doing that I would love not to have to do.”

In contrast, radio-frequency-mapping company HawkEye 360 and Internet of Things startup Kinéis chose to outsource their constellations, but not without difficulty.

“The bottleneck isn’t launch right now, frankly, it’s the manufacturing capacity here in the United States,” John Serafini, chief executive of HawkEye 360, said at a U.S. Chamber of Commerce event in December. “American manufacturers of microsatellite buses [are] relatively paltry. Most of them are [research and development] type facilities. There aren’t many set up at scale.”

SFL is currently building HawkEye 360’s next 15 radio-frequency-mapping cubesats, which the Herndon, Virginia, firm expects to launch by the end of 2021.

Kinéis, which is based near Toulouse, France, chose bus supplier Hemeria and payload provider Thales Alenia Space to build 25 cubesats partly because of their local proximity. “They are about a 20-minutes drive from our headquarters,” CEO Alexandre Tisserant said. Plus, Hemeria and Thales Alenia Space have built earlier Argos location and data-collection hardware that is the basis for Kinéis’ constellation. Kinéis employees also have more experience with designing applications than building satellites, Tisserant said.

“Maybe on paper it would be economically more viable to do it by ourselves, but with the state of the knowledge and the people we can find and we have in the area of [Toulouse] France, I think this is definitely the best choice.” he added.

There is a healthy ecosystem for the new actors building one or two cubesats, said Chris Boshuizen, a Planet co-founder who now works as an operating partner for Data Collective, a San Francisco venture capital firm. “Multiple vendors have similar products with varying price levels and features,” he added. For new companies, “buying stuff is absolutely the right idea” because “it will get you started quickly.”

The calculation is different, though, for large constellation operators. They have to look at the cost of every component from radios and reaction wheels to thrusters and processors. If the constellation operator has the in-house expertise to build satellites with more capability at a lower cost than what it can buy, it makes sense to vertically integrate, Boshuizen said.

For now, cubesat manufacturing is “a healthy boutique market, not a thriving large-scale manufacturing industry,” Boshuizen said.

The question is how will it become a large-scale manufacturing industry unless companies place large orders.

This article originally appeared in the Feb. 24, 2020 issue of SpaceNews magazine.