Before Elon Musk started SpaceX, he held a series of salons in Los Angeles and Silicon Valley, asking experts in the field for ideas on what a new-age rocket company should look like. A handful of people advising Musk urged him to make small, cheap rockets that would bring the cost of getting something into orbit down to $5 million from the going rate of about $100 million. In the years that followed, Space Exploration Technologies Corp. did manage to lower launch prices drastically, but it stuck to making bigger spaceships instead of pursuing the radical approach that some of Musk’s advisers desired.

A couple of these space dreamers, who helped get SpaceX off the ground, have now turned up with their own rocket startup called Vector Space Systems. They have radical plans and then some. Vector’s goal is to make a $1.5 million rocket that can carry small satellites into orbit. It expects to conduct 100 launches per year—a figure that would match the annual capacity of the entire aerospace industry.

To help meet this goal, Vector told Bloomberg that it raised $21 million from some of Silicon Valley’s largest venture capital firms, including Sequoia Capital, Shasta Ventures and Lightspeed Venture Partners. “While Elon made space safe for investors, he did not fix the launch problem,” said Jim Cantrell, co-founder and chief executive officer of Vector. “We are making the rockets as simple as they can be made, building them like sausages and launching by the hundreds.”

Cantrell is something of a connector in the aerospace world. He’s spent decades helping companies arrange deals for launches and assessing the business plans of startups. Most famously, Cantrell accompanied Musk on a handful of trips to Russia in a bid to negotiate the purchase of intercontinental ballistic missiles that Musk hoped to use as rockets. When those negotiations fell apart, Musk set off to start SpaceX in 2002 and build his own rockets.

One of the rocket builders Musk visited in Los Angeles during the early years of SpaceX was John Garvey, a former engineer at manufacturer McDonnell Douglas. Garvey had long fantasized about making very small, very cheap rockets and maintained a workshop to pursue his designs. He’s now chief technology officer at Vector, and Cantrell has helped find the money to turn his prototypes into a commercial business.

Source: Vector

SpaceX has become the low-cost provider in the rocket-launch business and changed the economics of getting to space through its reusable rockets, modern electronics and breakthroughs in materials science. Its workhorse is the Falcon 9 rocket, which stands 230 feet tall and can take 50,000 pounds of gear to low-Earth orbit. Typically, the Falcon 9 will carry a bus-sized satellite or ferry supplies to the International Space Station for about $60 million a pop—about a fifth of what its rival United Launch Alliance LLC charges.

The rockets Vector has started building in a factory in Tucson, Arizona, are tiny compared with the Falcon 9’s. The company’s first spacecraft, the Vector-R, stands just 42 feet high and can take only 132 pounds of stuff to orbit at a cost of $1.5 million per flight. This means Vector cannot address the bulk of the launch market, which centers on sending satellites that weigh thousands of pounds up for commercial and government customers.

Vector, though, is betting on the coming wave of smaller, cheaper satellites known as CubeSats to be the core of its business. These devices are about the size of a shoebox and have just started becoming popular in recent years as improvements in electronics and software have made it possible to shrink satellites drastically. Planet Labs Inc., a startup in San Francisco, is the best-known CubeSat maker. It has a constellation of 160 satellites circling the Earth, taking daily pictures of the planet’s landmass. Other small satellite makers are rushing to market with machines that take pictures, perform communications and conduct experiments in zero gravity.

Vector has entered a suddenly crowded field. Several rocket-launch companies have appeared in the U.S. and overseas to serve as ferries for small satellites. The most successful to date is Rocket Lab, which just completed its first test launch from New Zealand. Its rockets can carry more weight than Vector’s and cost $5 million per flight.

The top performers in the launch business, like SpaceX and Europe’s Arianespace SA, conduct about one liftoff per month. Both Rocket Lab and Vector want to ratchet up this cadence. “We think the market will be on the order of 400 to 500 launches a year,” said Cantrell. “There is room for four or five launch providers.”

Source: Vector

Vector, which started operations 18 months ago, has been testing early versions of its rockets in the Mojave Desert, although it has not yet tried to reach orbit. About eight small satellites could fit inside of a Vector-R rocket and close to 30 could fit in the larger Vector-H rocket that is due to be built over the next 18 months, according to Cantrell.  The company will have its next test flight in late July or early August and aims to be flying commercially by mid-2018.

Rocket companies run notoriously behind schedule, and that launch date is very aggressive for such a young company. While Rocket Lab has a private spaceport, Vector will be looking to fly from government-run launch pads in Alaska, Florida and Georgia. “We will be able to launch in three to four hours from standing the rocket up, to fueling it, to gone,” said Cantrell.

The aerospace industry is divided on how successful these small rockets will be. The price per pound on the large rockets from SpaceX and others is still more economical. But it’s the flexibility of requesting a launch, almost like you’d order something on, that could end up being more attractive than pure cost. “The lower end of the market will be more important than most people realize,” said Rob Coneybeer, managing director at Shasta Ventures. “Moore’s Law is allowing you to make more capable things smaller and smaller, and I think the low-end rockets will hit the sweet spot.”

Read More: