Baiju Bhatt
Related people
Episodes
Notes
Grit · Data Centers in Orbit by 2028: Baiju Bhatt on His Plans After…
-
When Will Baiju Bhatt Launch an Orbital Data Center?
Baiju Bhatt says the goal is to launch the first in-house built system at the end of 2028. He does not think the opening will last forever, so the work has to land while the demand is there. He calls the date ambitious and says the team is working hard to hit it. An in-house system by the end of 2028 is still inside the time it takes to build a big data center on Earth, and shorter than a lot of the timelines he sees.
The rocket program exists to serve that data center in space, not to do human spaceflight and not to carry other people's payload. He says they may do more later, and that is not the plan now. They need a rocket program that works. They do not need the world's greatest one. On
-
What Did Robinhood Teach About Controlling Unit Economics?
Baiju Bhatt says a lesson from building Robinhood is that controlling the unit economics of what you sell leads you to vertically integrate, and to control the components that drive the cost. Robinhood did a lot of that. Asked whether a space company can borrow the stack that already flies and only redo part of it, he says a large chunk has to be redone.
There is nowhere near enough launch capacity for what everybody wants to do. There is not enough mass to orbit, not enough rideshares, and not enough ways of getting stuff to space. SpaceX's accomplishments are monumental, he says, and he does not want to speak for them, but they are going to use a lot of their payload for what they are bui
-
Why Did Baiju Bhatt Start a Space Company After 15 Years?
Baiju Bhatt took another swing because life is finite. Working on a company for 15 years means he is not the youngest founder anymore, and the years when he can give time and energy without much else in the way are not infinite. He remembers Robinhood as physically and mentally grueling. Being successful as an entrepreneur zaps every bit of energy you can give it. He wanted a real swing at a space company while he still felt young enough to grind hard.
He still looks for reasons to push. He says he is a bit like Michael Jordan, who would look for something to fire him up, and that an entrepreneur has to exert unnatural force to bring an idea to life. Clocking in nine to five will not do it,
-
How Do You Know You Have Product Market Fit?
Baiju Bhatt marks product market fit when results do better than the projection. Doing better than what was projected is usually a good sign of product market fit, and the comparison is the most general version of the test.
For a consumer product, product market fit is part and parcel with long-term retention. People continue to use the product time and time again, or the product dwindles off to zero. The hallmark, in all the times he has seen product market fit, is how results line up with projections for how well the product is going to do. Expectations come in three grades: mediocre, good, and great. Doing better than great expectations generally means product market fit. Something is go
-
Why Does a Space Company Need Factories?
Baiju Bhatt is building a rocket engine factory in Seattle because a space company has to build a lot of factories and other facilities. Factory building was not as big a part of Robinhood. He does not think Robinhood had factories for anything. Robinhood was a digital product.
The rocket team and the satellite engineering team are being built in Seattle, and the core of the company is still in Silicon Valley. Pretty much everybody building the first two satellites and the first two satellite missions is in the Bay Area. Long term, the computer science and computer engineering people probably stay in the Bay Area, along with a lot of the executive functions and the compute sale side of the
-
Why Will AI Agents Need Data Centers in Space?
Baiju Bhatt says data centers in space will get more useful as AI agents spread, because those agents need a tremendous amount of compute. You do not have to believe humanoid robots will take over. A more pedestrian case is enough: billions of people with AI agents working in the background, doing menial day-to-day tasks.
If every person could have an agent helper, not a person, handling those tasks, people would use it. He already hits the shortage. Claude Code times out because it is running out of compute, and OpenAI and Gemini run into the same wall. There is not enough compute to ask for a recipe for a Magic the Gathering deck.
-
How Much Will Data Centers in Orbit Cost by 2028?
Baiju Bhatt thinks it will take low single digit billions to get data centers into orbit by 2028. He announced a 275 Series B, had raised about 90 before that, and total capital raised is around 365.
Fundraising will be a big part of his life and a big part of the job. He got stuck with the emails and spreadsheets while engineers work in the clean room, building stuff and getting the software work done, and he spends the time on Zoom calls. The Zoom calls are what the job calls for. Fundraising is an important input, and a ton of business development goes into the work too. Beyond the space company work, he is learning a ton of new stuff and really liked it a lot.
-
Why Is Heat the Hardest Part of a Data Center in Orbit?
Baiju Bhatt says dissipating heat is the thorniest part of a data center in orbit. In broad strokes, the satellite uses solar panels to generate power, GPUs for compute, supporting electronics, and a big heatsink that absorbs heat and then radiates it away. Large radiator sheets mean carting a bunch of mass and volume to space.
Two racks of GPUs are strictly worse when they are split apart, one in one room and another in a building across the street, than when they sit together. Two racks should be one big satellite rather than two smaller ones. How much radiator mass has to fly is the constraint that design has to beat.
-
Why Put a Data Center in Space Instead of Beaming the Power Down?
Baiju Bhatt says the power should be used in space instead of beamed to Earth, because transmission and the conversion from electrical to optical and back throw away a lot of energy. Harnessing the power in space and using it there leaves more energy to work with, and from a first-principles standpoint that should produce better economics than beaming it to the ground.
Layering compute on a commoditized service like energy brings more margin on the way to break-even. The path he landed on was space solar power plus compute: data centers in space, without the power-beaming part of space solar. Toward the end of last year and the beginning of this year, the plan locked onto the unit economics
-
How Did Baiju Bhatt Start His Space Company?
Baiju Bhatt is solo founding his space company and wrote the first check himself. His time is extraordinarily valuable, so putting that time in without putting his money in does not compute, and he figures he probably has less time than money.
He did not consciously look for a co-founder. A very strong force was pushing him to build a space company, and he wanted to try it himself. Older now and used to operating businesses, a co-founder is harder to bring in, and the work is a lot lonelier. He is putting his time into the company and seeded it with the first check, on the order of 10 million bucks.
The balance he keeps is time versus resources. With less time than money, the money goes in
-
What Is the Core Idea of the Case for Mars?
Baiju Bhatt says the core thesis of The Case for Mars is an engineering approach to a Mars trip: look for reusable components, bring cost down, and build closer to what the hardware actually costs instead of what the government spends. A scientist who had worked on the government-style plan wrote the book to argue that anyone who actually wanted to reach Mars would not do the trip the usual way. Reading it made him want to try the same approach in his own career if it was possible.
While he was building Robinhood, SpaceX landed a reusable rocket for the first time. He felt that if the landing was happening in his lifetime and he was not part of it, he would look back as an old man and funda
-
Why Is a Rocket's Upper Stage the Data Center?
Baiju Bhatt's architecture makes the upper stage of the rocket itself the data center. The metal already on the vehicle is the heatsink, so the stage is kept and put to work instead of being thrown away or only reused as a launcher.
Robinhood's thesis was that mobile was an emerging platform. His bet is that low-Earth orbit is a platform for commercial activity, and the problem was how to make something from space that can compete with doing the same job on Earth. He looked harder at powering data centers once it was clear that was where the energy use was going. A rocket on the way to space already carries a bunch of metal. Using that metal as the heatsink, joined to data-center power dema