Ronak Kumar Samantray, founder of TakeMe2Space, didn’t chase investors or flashy press. “Reddit is where we found most of our customers, because that’s where all the geeks are,” says the head of the Hyderabad-based space startup that offers compute in orbit at USD 2 a minute. Speaking to TNIE as part of Hyderabad Dialogues, he talks passionately about building a satellite at his home in Gandipet, setting up data centres and how easy it was working with ISRO thanks to IN-SPACe. He also shares advice for those looking to build a startup. Excerpts from the interview:
How did you get into satellites and space?
My first startup was acquired by Reliance in 2019. I was then looking at what to do. Space caught my attention because I believe humans are definitely going to become a space-faring species. Back in the day, launch costs used to be USD 1 million per kg. They came down to USD 45,000 per kg with Falcon 1, then to USD 6,600 per kg with Falcon 9. It is projected to go down to USD 600 per kg with Starship. Since rocket costs are reducing so much, I was keen to see what we can do with satellites.
Around 95% of the satellite industry is earth observation: you put cameras in orbit, take pictures and download them. And you have communication, like our INSAT satellites or Starlink. And the customers mostly are defence, the GIS industry and so on. So I wanted to see what we could do.
And I have a bias towards computer science because I've been a coder throughout. When I was looking at a satellite, it looked like a computer. It has a computing capacity, power generating capacity and it has certain sensors around it. So if you could put an array of computers in space at a very low cost, what are the new business models that can evolve?
As an engineer, I'd made some capital, I know how to write code. The core IP needs to be built with my own hands. I don't like working with somebody else or outsourcing it and then building their IP. So I started building a satellite, literally everything we did was built from my house in Gandipet. For satellite assembly, there's a clean room in the house.
We saw your daughter helping you.
Yes, that's why my satellite is called MOI-1, MOI-2. I asked ChatGPT: “My daughter is Moira, this is what I'm doing, what should my satellite be called?” Hence the name ‘My Orbital Infrastructure’ sort of emerged.
What goes into building a satellite? You have so many components.
True. I definitely did not know the complexity of what I was getting into. As an entrepreneur, you have to be stupid. If you're very knowledgeable, you won't attempt such a problem statement. None of us have been to space ever. When you read a research paper, let's say, orientation control of a satellite, how will you tilt the satellite? There are thousands of research papers and nobody tells you what the right answer is. Now, after four years, we are at a stage where we have indigenised all satellite hardware. That is exactly what drives my revenue today. All the satellite components that we have developed, that is what other satellite companies buy from us.
You don't import anything from outside?
What we haven't built are solar cells, propulsion and camera. Except for those, every satellite subsystem is built by us, designed by us, manufactured in Hyderabad. And that becomes a differentiator. Now, because I'm not buying anything, I'm not paying anybody else. I'm vertically integrated, the price is in my control. That benefit goes to the customer.
What caught my eye was that you said for USD 2, you can get something. Is it feasible?
Yes, USD 2 a minute. One, the launch costs have come down. Two, all the systems powering us are our own. The first satellite they were putting in, if you go ahead and buy that, it will cost Rs 3 crore. For me, it cost Rs 50 lakh, out of which Rs 25 lakh is just for solar cells. The rest Rs 25 lakh, everything is built in-house. So that is how much cost reduction we have achieved.
To make the entire satellite, it will be less than Rs 1 crore?
Rs 50 lakh, that's it. That's our biggest advantage. To launch a satellite, we are spending Rs 2 crore. Today, if you want to use a satellite to click a picture of a volcanic eruption, or the eye of the storm, you cannot. Even if you pay money, you are just trying to click one picture. You can buy a picture from a satellite operator. But nobody is going to give you access to a satellite. So, I am basically selling a printing machine for time. You decide what you want to print. Pay me USD 2 per minute.
What about data, images, resolution? Can you also elaborate on how it can be useful for a student, an amateur astronomer and so on.
Great question. Because the privacy laws come into the picture. We operate at 9 metres per pixel, a resolution that does not compromise the privacy laws of any nation. You can't see humans, trucks and all that.
Today, there is one common trend — AI. That is changing the way people do things in life. And that's why we are an AI lab. Say, if you are a researcher or student, or from the GIS industry. You want to analyse a certain area of land. Now, what do you do? You have to buy it. If they are the latest pictures, you have to pay. And everybody charges per sq km. Now, imagine you have to do it every day.
When you have to analyse how much the forest has grown over time, why do you need pictures for that? You can just download the vegetation index. That is the disruption we are bringing. We are not an Earth observation company. Our satellites are not optimised to download images. My downlink bandwidth is very small. My download speed is very small. Because what I expect my customers to focus on is that you capture the data using the camera. Do the processing. Only download the results. Then the cost gets optimised. And the cost of inferencing is also low. Because today, what happens is when you are downloading, let's say, 1 picture is 1 GB. When you are downloading terabytes of data, somebody is paying for the cost. Then somebody is paying for the analysis. And then you get the results. So the end customer has to pay for the entire value chain. But in our case, what we are saying is, as a customer, just download the results. And pay me for the time.
How does someone communicate and ask it to do data processing?
We have a web dashboard. You sort of upload your area of interest, say, Hyderabad. You have to write your code. We want to target people who know how to code first. You write the AI application which can take the image and do image processing. Upload the AI model. And say submit. Behind the scenes, we run it on a digital twin. And then we upload it via the ground station to the satellite. And the satellite is configured that at this timestamp, at this latitude, longitude, trigger this application, do the processing and download the data. So the entire uplinking, satellite tasking and downlinking are what are automated. The user doesn't have to learn all these steps. And there is no human intervention either.
So using it requires knowledge of coding. Have amateurs used it?
When we did the MOI-TD mission, one of our customers was a Malaysian university. But guess who the second customer was? A Class 9 student from Chandigarh. They wrote a basic code. They wanted to figure out how much UV radiation hits our upper atmosphere. That's a basic code. That code is something every child knows today.
Then there were school students from Rajasthan. They wanted to monitor how the temperature changes. Satellites give a very interesting profile. This is Earth. One orbit is 90 minutes. So, 60 minutes you have sunlight, 30 minutes of darkness. So when you are in sunlight, you are at 85 degrees. When you are in darkness, you are suddenly at minus 60. That is how drastically the temperature changes.
How did they know about your startup?
It's an organic thing. For me, most of the marketing happens through Reddit. Because that's where all the geeks are. Not on Instagram. Not on Facebook. I would keep posting about what I am working on. You look at our LinkedIn page, it will be mostly about our work. We even post all the failed experiments on LinkedIn. Reddit is where we found all our customers.
Is it feasible to build data centres in space?
The temperatures here have been going up. The rainy season is unpredictable. Whether you build a data centre on earth or space, the cost of the hardware is the same. But the cost of the environment is different. Space gives you unlimited free energy for a lifetime. There is no pollution. There are no humans throwing stones on your solar panel. There is no maintenance cost for your solar panels. You get 30% more solar energy in orbit than on earth because of the atmospheric filtering. Another issue is thermal cooling. Data centres spit out a lot of heat as well. Cost of running air conditioning is going up. Space as an environment takes care of the thermal cooling. There is no cost of cooling. If you could put an array of computers in orbit at a lower capex, then space as an environment gives you a lot of advantage in the operational costs.
Rocket guys are taking care of the launch costs. Our job is to ensure that satellites are cheaper. And that is what the long-term play is. Whenever the industry gets to USD 600 or USD 100 per kg, we should be the first ones to set up a data centre in orbit. Today, the industry is at USD 6,000–10,000 per kg. It is not viable. That's why it is USD 2 per minute. Our aim is to get to USD 0.002 per minute.
You have already sent two experimental satellites to space. How was it working with ISRO?
It was brilliant. IN-SPACe is the arm of ISRO that helps you liaise with the space organisation. So they make it super easy. They're super helpful, because they have been doing it for the last 30 years. They want to give you knowledge. The government has created a framework. Without IN-SPACe, it would have been a nightmare.
How many people do you have on your team? Initially, you said you didn't know much about the field. But you had to convince the right people to join you. Apart from the capital, how did you persuade them?
We have around 20 people. No, I never show capital. Because the moment capital gets involved, the nature of humans changes. What has worked for me is whenever I meet somebody, I will tell them my entire idea, this is what I'm working on, this is what my future strategy is. So, I try to monetise trust, I put the trust on the table first, and you hope that the other person will also trust you. Be transparent to people, and then people sort of start trusting you. Of course, I'm selling you a dream, but I'm also backing it up with my work.
Some of their specialisations?
My head of products, Sharat, he's a super cool guy in mechanical engineering. When he designs an object, he thinks about manufacturing things, about assembly, disassembly, takes everything into consideration when he designs a product. There's Firoz. He has been a guy in the Hyderabad ecosystem who's known for SPM, special purpose machines. People would call him when nobody else can build that machine or fix that machine. He has been a freelancer since after college. He has a huge vendor network access.
There's Anand, who was the CEO of T-Works. He knows how to work with the government. He has run large organisations. So he heads our business. Now we have a physical sciences team. There’s a team member who’s exceptionally good at maths. There's a PhD guy who was working on black holes. My entire avionics, my circuit is designed by a 22-year-old boy. Because he's super passionate. Most of us have no experience in the work we are doing today, but all of us have a passion to do what we're doing today. Our passion backs it up.
Suppose a youngster, just graduating out of say an IIT or some other university, wants to build a startup. What would you tell him?
Just do it. Because it's only when you do it, you figure out the reality. There are two reasons to build a startup. You should be very clear — do I want to make money or am I doing it to change the world? Because they're two different tracks of execution. If you want to create something that no one else has done, then you have to look at micro trends that nobody else understands today. And that is what my approach is. How do I validate my idea? When I was working on this whole technical space, I would go to 10 people. If nine people say this is nonsense, I said, okay, I'm on the right track.
Whereas, if you want to make money today, nine people have to say, yes, I would love to use your product, right? So you should be clear on what you want to do. For example, today, quick commerce is killing it in India. So if you want to do something today, please get into quick commerce. That's where the money is today. Second, once you have figured out the problem statement, figure out what the kernel is, like computer science. Build it yourself, never outsource.
If you could go back to when you started TakeMe2Space, what would you do differently?
Given a choice, I'll never start a company. I'll chill in life. See, as an entrepreneur, I never advise anybody to start a company. It's a nightmare. You're cutting the same branch that you're sitting on. But like I said, once a psychopath, always a psychopath.
TNIE team: Kalyan Tholeti, Prasanna RS, Vennapusala Ramya, Nitika Krishna, Darshita Jain
Photographers: Vinay Madapu, Sri Loganathan Velmurugan