Why Improbable had to build Somnia, its 1 million TPS EVM blockchain

In the latest episode of the Blockchain Gaming World podcast, editor-in-chief Jon Jordan talks to Improbable CEO Herman Narula about the history of the company, its involvement with Yuga Labs’ Otherside metaverse, and why he felt Improbable had to create and launch its own blockchain.
BlockchainGamer.biz: Can you give us some background about Improbable?
Herman Narula: Improbable has had a master vision from the time when we started as a bunch of computer scientists out of Cambridge to build the infrastructure for virtual worlds and for what we think of as the new economy, the future economy.
I talk a lot about this in my book, Virtual Society. I think the new economy is about a couple of things. It’s about unbelievable experiences that we all value, whether that’s going beyond games into massive battles, or being able to live in living autonomous worlds, or being able to interact with people that you might never get to meet in the real world, but you can engage with it in an amazing sense in virtual reality. That experience is one big part, and boy has it been tough to build the technology. But now we’re at the point where I think that’s in a fairly established place.
The second part is the tools so that anyone can build these experiences and so that they can be accessible to people, to companies and to organizations that might never have considered the ability to build a virtual world. It is really complicated to build, operate and run a complex online game. And I think we’ve made that a lot easier with some of the companies and products that we’ve built.
The third is around the economy. That’s what took us to blockchain. We had been waiting for somebody else to solve this problem but we realized no one is. And so we had to. For these worlds to really come to life, they need roots that extend into the real economy and extend into our culture in order to feed these worlds and to imbue them with meaning. That means having the infrastructure for people to own things and for those things to be able to transfer easily between worlds, which I think is a big part of what we’ve built around MSquared.
The technology is hard to build, but looking at Improbable’s history it seems that getting the business model right and gamer culture are more difficult?
It was really tough to break into the games industry. We finally found a lot of success with The Multiplayer Guys, which was a company we acquired and then grew its revenue 10x. It worked on Call of Duty. It worked on Grand Theft Auto. It worked on loads of titles and even built the multiplayer mode for Fall Guys. In doing all of that, we gained a tremendous amount of experience, but also respect within the games industry that fed to the second version of our core technology and the creation of MSquared pretty profoundly, moving on from where we were with SpatialOS [Improbable’s first product].
The most interesting thing about it, probably beyond the deep technology venture builder that we’ve become, is the tenacity with which we’ve pursued the mission. A lot of people give up if it doesn’t work out immediately or if the problem is really hard. I think we’ve specialized in doing things that everybody tells us we can’t do, and doing them in a really profound way.

Not many people know this, for example, but myself and my co-founders were sanctioned by the nation of Russia. We’re not allowed to travel to Russia or Iran or North Korea or any these places. Not that we want to go to North Korea. It’s because on the journey with our technology, we applied it to defense work. We were able to attack yet another very hard problem and got involved in the Ukraine war. We actually sold our defense business in 2023 but it goes to show the character of the company. We go after things everyone tells us we can’t do, which is why we’re building a blockchain, right? Because that’s the thing everyone really doesn’t want you to do.
Is this a case of the vision being correct but adoption taking much longer than expected?
One of things we’ve brutally focused on with MSquared, which is our metaverse business and the manifestation of why we originally started Improbable for the experience-side, is making the cost of operations so low and the ease of use so high in terms of the platform. We’ve done over a hundred public events in the past year from major league baseball matches to K-pop album launch parties. We’ve done this to perfect the capability.
Last Saturday, we broke the world record for the most people in a first-person shooter in a single space with over 1,500 people. But yes, you’re right. Commercialization and timing are very challenging, which is why, as we’ve evolved into a deep tech venture builder and why we’ve become a venture builder is realizing that each piece we build can be valuable in its own right. Also some of the skills and technology we’ve developed are super useful to other companies that we might engage with and help them to grow. A lot of that is commercial timing. A lot of what we’ve brought to the companies we work with is this idea of how to commercialize the capability that they’ve created.
What do you mean that you’re a venture builder?
Improbable is a company of companies. We take each of the things that we invent or the people that we engage with and want to build companies with, and we turn those into individual companies. So MSquared, run by Rob Whitehead, my co-founder at Improbable, is actually a business with its own cap table, its own board, that is a subsidiary of Improbable. There are many other ventures we’ve built including things like Jitter, which is our Twitch play, allowing people to interact directly in Twitch chat with their audiences, which is really awesome.
We do it this way because when businesses try to compound everything together, you end up creating a very unfocused machine that isn’t capable of being effective at any one thing. Separating things into individual businesses means we can take our large cash pile and potentially engage with external parties to work on cool ideas that we think matter. But it also means each thing we do tends to be very focused and can either succeed or fail. I think what’s been great about our model is we’re also happy to accept when things don’t work and to be able to move away from them.
This meant that as we’ve worked toward building a blockchain with Somnia and the Virtual Society Foundation, that’s been really focused. We don’t need the Somnia blockchain to be deeply integrated if it doesn’t make sense to be with our other ventures. It has to stand on its own two legs and that has been a very effective discipline for us.
Can you talk more about MSquared?
MSquared is a software platform for building a virtual world. You pay a little bit of money every month and you can build and operate a large-scale complex virtual world just like Yuga Labs’ Otherside. It can handle almost 30,000 people in a single space interacting together. It allows for streaming. It allows for crowd audio in the same spot so everyone can speak and talk and interact together. You can completely customize a game world with the full power of the Unreal Engine if you wish and have that work with thousands of people. It handles 20 billion messages a second on the back end and it’s so cheap and low cost that an indie studio could do this.
For those who have been following Improbable for a long time, this is what we wanted to build with SpatialOS 12 years ago, but it took that long to get to a place where this works as effectively as it does. It’s growing pretty quickly and we look at it as one of the most important businesses that we are enabling.
Why has it been so hard to build?
We had to invent our own rendering engine to handle that many different looking characters on screen. We had to invent a cross-world format so that people could bring assets from one world to the other. We had to invent bandwidth compression technology. We had to invent a backend that could handle that level of scale. It took us a long time to do that and many false starts in terms of getting this to a production-ready state.
People have asked me, ‘Why has no one else been able to build that level of scale?’ Why can’t you do it in Fortnite? Why can’t you do it in these places? Because we’ve done nothing but focus on this problem for an extended period.
Now a lot of NFT collections and game communities need to differentiate and provide social spaces and interaction. MSquared is perfectly positioned to do that. And on top of MSquared, we’ve also built a protocol – we haven’t talked very publicly about this, but it’s in the manifesto that we released – it’s a way for worlds onchain to have shared economies and shared asset networks that can move between each other.
A lot of people have asked ‘When will Bored Apes move into another world?’ Well, we did that last year with Major League Baseball allowing apes to come into those worlds and those assets being part of different universes.
Isn’t one of the issues with these large scale social spaces that most projects don’t need to have 30,000 users in a concurrent area?
For some projects it’s hard to even get a hundred, so what we’ve done is taken away that scale problem completely. We’ve made it really easy to build and operate a world of arbitrary scale, that’s the way I would look at it.
When you’re designing or thinking about your game or world or experience, you just don’t have to think about scale. For a lot of concepts, a lot of ideas, social communities coming together, worlds that you can interact with, especially for celebrity interactions, some of the biggest areas of growth for us are sports and music, where if you’re a musician or a sports star wanting to hang out with your fans, you can’t shard yourself.
So a lot of the value that people have often talked about with the metaverse just doesn’t work. I mean if you saw VR-based music experiences, they didn’t work because you were dancing by yourself with five people around you, right? With what we’ve built now, suddenly you can have that crowd, you can have that experience. One thing we’ve always bet on is the need for human beings to feel related to one another. Having an unbounded ability to bring people together is always a good bet to make.
Even in the games industry, it’s been a really rough time. 10,000 layoffs this year, last year, the year before. The games industry desperately needs differentiated content. When we started Improbable, it didn’t. Now it really does. And so if all you’re going to do is come out with yet another arena shooter or yet another experience that looks just like everything else, your game is going to get crushed. Suddenly, it’s very important to have differentiation. I’d also say it’s not just about people anymore. It’s about AI too. The platform is built from the idea that you’re going to have thousands of NPCs interacting potentially as well, which kind of changes a lot of what people want to build.
Which brings us to Somnia. Why does the world need another blockchain?
The challenge is there’s block space available, but I need sub-second latency and I’d like more than five transactions a second. The reality is that if you want to build a serious business on top of crypto, you really can’t. If you want to trade memecoins or you want to launch an NFT collection, you’re spoken for. But if you want to build an onchain world that everyone can extend, the mess, particularly for EVM.
The reason we came to blockchain is that we wanted to put much more of our worlds onchain. And we found that the cost of doing so would be astronomical. So we thought we may as well build our own. We’ve been optimizing, optimizing, optimizing. Now we’ve reached the point where we have testnet live this week, and we’ll be able to handle a million transactions a second, 50,000 Uniswaps a second in one of the experiments that we did. We can handle non-parallel transactions, sub-second latency, validated nodes that are less intense than Solana so you can use consumer hardware, and an ability to be completely EVM compatible.
This is going to be profoundly valuable to games companies and other organizations that don’t want to worry about gas fees or block price and want to be in a place where this stuff is just ridiculously straightforward to use.
We didn’t want to do this. We have enough going on, but the fact that nobody had done it after 10 years of waiting is what pushed us to the point where we just need to move forward and make this happen. And it’s not just a blockchain. It’s also a governance unit and in a way of operating where you have Improbable and an organization that we’ve supported in the Virtual Society Foundation as a counterparty. So if you’re a serious business looking to build on top, here’s a multi-billion dollar company that’s actively able to support this. Also there’s someone to sue if things go wrong. There’s an opportunity to really engage.
So you’re not going to optimize for the Somnia token price then?
Honestly, no, because we don’t need the money. I mean, that’s the reality. We’re investing significant cash every year. It’s public record that we have almost a quarter of a billion dollars in cash. We’re in a position where our focus is building the best infrastructure and creating the best environment for developers. It’d be great to have a token price that actually incentivized people to come in to get involved in the community. And that’s what we’re hoping the foundation that’s independent of us does.
The other thing you mentioned earlier is we’ve been doing this for 12 years, right? A lot of the people who enter the blockchain space to “change the world” have been doing it for six months. They cash out and they run away. Hopefully we’ve developed enough trust in the community that there’s a level of tenaciousness to what we do, we’re not going to do something and then run away from it six months later.
Is there some threshold at which the speed of the blockchain doesn’t matter anymore?
It’s like dial up and broadband, right? We weren’t doing stuff with dial up that we could do with broadband, but you things to be at a point where you’re not thinking about that capacity. When we talk about onchain games now, nothing’s onchain, right? There’s a token, but the world state isn’t onchain and an NFT is just a pointer to an image. What’s an example in your mind of a fully onchain game? I’d love to challenge you on this.
The one I work with is a football simulation called Soccerverse, but I don’t think you would have heard of that, so maybe Pirate Nation is the best example?
Pirate Nation is a great example. Its infrastructure costs are absolutely astronomical because of the problem of performance. If a game like Pirate Nation was running on Somnia, it would save millions of dollars just because the gas fees would be lower
That’s a great example for me because Pirate Nation shouldn’t have to spend that kind of money. It shouldn’t have to spend the sort of money you would spend to run a complex 3D MMORPG just to put a relatively simplistic game state onchain.
People have made this happen despite the enormous challenges of doing this but I want to be in a place where there’s not any thought process putting stuff onchain. Then the crucial thing is shared state and context. Much of the point of blockchain is the idea that I can build off of your work and add value to it. How am I going to do that if your state is on your own app chain and my state’s on my own app chain? At this point, they’re just databases, right? It’s not a blockchain anymore. It’s a series of gated communities, which is part of the problem.
What we’re aiming to do is not just handle performance, but also bring everything together. It’s not going to happen overnight. It’s going to be a long, slow grinding process. And I think what we’ve shown in the past and what we hope to show again is that we can outlast a lot of these problems. We’re planning a five-year investment.
Who’s your ideal customer for Somnia?
I don’t think our focus is going to be getting the stuff that’s already onchain to come onto Somnia. Our focus is, as always, new businesses, new opportunities. If I think of some of the people we’re speaking to about this, they’re some of the largest consumer brands on earth, and they’re not currently selling NFT collections. They’re thinking about wholly new experiences onchain. Of course, every world being built on MSquared, which wants to put something onchain, we’re going to make it really attractive to do that as well.
When I was told blockchain would be a thing, like you can mod my game, you can build this experience, none of that is possible today. It’s a very insular community focused on upping NFT value. I don’t particularly think there’s a lot of value in servicing that. That’s very much become PVP. Folks giving out massive grants to try to move projects onto their block space when all that block space is fundamentally the same.
With the performance we’ve reached, you could take the entirety of every single transaction currently happening on blockchain, move it onto Somnia, and still have room left over.
How does blockchain improve the game industry?
Games struggle enormously with two problems. The cost of user acquisition, which is almost $30 or $40 a user, whether you’re a blockchain game or not. And also the massive cost of developing new experiences. Unless you share context and share space, you’re never going to take advantage of the advantages of blockchain, which allow you to reduce your user acquisition costs, by doing things such as airdropping value to popular players of another game, or even being a place where you can create cross-chain value. If you can’t do that, you’re just making games with tokens.
And all that means is your game isn’t going to be on Steam. You have no advantages beyond that. I think the real reason people are doing this is because they want to pump a token and they know that an infrastructure token is more valuable than a non-infrastructure token. The market will only bear a certain number more of those grifts before there’s a real opportunity to ‘come to Jesus’ and fix this up. We’re very much playing for the long term.
So you’re not optimizing token price?
I couldn’t care less what our token price is. It doesn’t matter at all. That’s not how we make money. We’re going to succeed based upon the amount of transactions, traffic and users, so all I care about are the number of developers and the number of users. To the extent that token price in any way influences these, I care. But that’s it. It truly does not matter.
I don’t know when we got into this habit as an industry that the roads and the power plants should be generating value through their speculative value. This is what’s wrong with crypto. This isn’t something we think about on a daily basis. We do worry about how to make it really easy for developers to come onchain. We do worry about expanding the pool of creatives that can come onto the blockchain. A lot of what I’m trying to do with our venture builder is apply millions of dollars of investment capital into compelling businesses that we think might benefit from being onchain.
Why is the performance of current blockchains such a problem?
Again, I can’t stress enough how the tech not being there is an issue. You couldn’t put Call of Duty’s economy onchain. It would blow up every blockchain you launched. We keep hand waving over that as an industry because we’ve been fed this idea that it can’t be optimized further. It’s shocking to me the level of optimization on EVM that’s been possible. We had to build our own byte code compiler. We had to come up with really complex ways of doing hardware parallelization, invent our own database. But we’ve got experience doing this before.

And the current blockchain games sector is like the Hunger Games. There’s a lot of people fighting over a very, very small amount of value. That’s because they’re in it for their own personal exits and their own personal opportunities. I can’t blame them. If this is your first venture and if you can be an anonymous person online and exit with 10 or 20 million dollars of cash personally, more power to you.
But ultimately that isn’t how we’re going to make a multi-hundred billion dollar industry, which is what crypto gaming should be, particularly in a world where the EAs are actually down in terms of revenue. So is Ubisoft. Activision post-acquisition is not in the best shape. And they’re down because gamers are bored of the same crap. We want new experiences. We want an opportunity to really invest in these worlds. If your choices are come play a play-to-earn game that isn’t a very good game and is mostly being played by a small number of southeast Asian folks who are trying to earn an income from it or come play Call of Duty 93? That’s a very sorry state for us to be in as an industry.
We absolutely would love to be responsible for a little bit of change in that equation. If we can attract the right developers and give them support beyond money, maybe we’ll be able to do that.
How do you think AI Agents change the industry?
We use AI in everything we do at this point, from changing how developers operate to building our technology. And now we’re thinking ‘What tools are best for AI?’ AIs have almost become the developer now in the way we operate and the way we think, even in terms of how our rendering algorithms work. It’s very focused on allowing UGC-made assets that are uploaded on the fly.
Our sense is that almost all of the labor that goes into game development will be automated in 36 months, that’s my view. I think it happens that fast. You will still have small teams that are really productive and necessary to correct and manage work. But in my own coding, I pretty much entirely use Cursor and everything is Vibe based. Of course, I don’t ship production code, but it’s kind of extraordinary how productive you can be.
In terms of why this matters from a blockchain perspective, look, of those million transactions a second on Somnia, how many are going to be generated by an AI? Probably a large number.
And if we’re going to be in a world where AI agents are doing commerce, promoting a game, building a game, they’re not going to have bank accounts, right? They’re going to have crypto wallets. Some people ask us about the capacity, ‘Are you building capacity that is necessary for the developer community?’ We’re like, ‘Sort of but it’s definitely necessary for an AI agent community’. I think this is where we are in today’s world. You have to be thinking about infrastructure for these two communities.
How significantly do you think that this AI wave will impact human behavior and jobs?
This is going to sound crazy, but it’s going to make online games into a critical industry for human health. Fulfillment is the lifeblood of why we get up in the morning and what we do. If you take away the fulfillment people get from their jobs – which is going to happen, it is an inevitability – then where’s that fulfillment going to come from? It’s going to be rich, meaningful worlds, which we can explore, which we can engage in, which we can see and create stories in. It’s so important for us to have purpose. For me personally, virtual worlds are more exciting than the real world.
I was looking at some stats yesterday. We were talking to a business that’s working in the space for our venture model that’s working on AI companions for relationships. I was shocked to see the stats. Some very not insignificant percentage of all US adults would be willing to have a relationship with an AI. The other side of this equation is if it’s good enough to marry, it’s probably good enough to replace your job.
Coming back to Q1 2025, what’s on the roadmap for Somnia?
The testnet is now live. I can’t say when mainnet will be but we are not just launching the chain and the token with the Virtual Society Foundation, we’re also going to be investing in companies on Somnia as well. We’re going to be bringing a large part of the capital at Improbable’s disposal to fully build out that ecosystem.
We’re really interested, especially in game developers. Bringing great non-crypto talent into crypto is the goal of Somnia. We don’t want to fight with the other piranhas for exactly the same five customers. We would rather do new things.
But presumably any EVM projects would be able to migrate to Somnia pretty easily?
It’s EVM compatible so you could change one line of the code – the endpoint. If you’re on EVM, literally press a button and we will just reduce your gas fees.
If Pirate Nation wanted to migrate tomorrow they could. I don’t know if they’d want to get off their own infrastructure for the sake of the token price, but if all they wanted to do is reduce their infrastructure cost by millions of dollars, we would do that.
The Somnia testnet is now live. Find out more via its website.
Comments are closed.