How Epic Games views the metaverse and the State of Unreal | Tim Sweeney

At its State of Unreal event last week, Epic Games showed off not only the visual magic of 3D graphics of games in the future. It also showed off the business models, the trends like user-generated content, and its Verse programming language.

Importantly, Tim Sweeney, CEO of Epic Games, showed how to think about the hype cycle around the idea of the metaverse.

During the event at the Game Developers Conference, Sweeney demonstrated how the company could maneuver through the minute details of Unreal’s substate shading system. And they talked about how Fortnite is integrating Unreal Engine (UEFN) tools so that everyone can make games.

During the event, Sweeney and his team showed off Unreal Engine 5.2, and they announced that Epic would give 40% of the net revenue of Fortnite to creators. They showed the tech behind games like Senua’s Saga: Hellblade II and the Metahuman Animator, a tool that can be used to create extremely realistic animations, based on video captured from human actors and almost instantly converted into an animated structure that can be used to create 3D animations for games and films.

After Epic’s State of Unreal was over, I interviewed Sweeney and Sax Persson, executive vice president of Epic Games, at Epic’s booth at the Game Developers Conference.

>>Follow VentureBeat’s ongoing GDC 2023 coverage<<

We covered a lot of ground, including topics like the open metaverse, how to defeat walled gardens, the role of UGC, the need for a metaverse programming language and how to create open standards for the future. And we addressed how to solve the problem of the “sniper and the metaverse“?

Here’s an edited transcript of our interview. During our session, Sweeney was devouring a big bag of popcorn.

GamesBeat: It’s a very interesting State of Unreal. You talked a lot more about the metaverse at the end there. Was there some intention in that?

Tim Sweeney: It needed to be explained. There’s been so much hype around particular agendas, like selling NFTs or pitching new products that haven’t found an audience yet. I just wanted to highlight how real this current trend is. There are hundreds of millions of people doing this already. There’s a path to doing it in an open way. It’ll be another one of the big failures of our generation if we let another new ecosystem develop into a set of walled gardens. It doesn’t have to be that way. In fact it’s probably better for all participants if it’s not that way.

GamesBeat: It also feels a bit like people got tired of the metaverse already. Now they’re excited about generative AI. The metaverse is yesterday’s hype. Before that it was blockchain. It sounds like you’re also not saying that these things are all just equivalent as hype cycles.

Sweeney: If you chart the trajectory of this thing over the past seven years, it’s hard to say that MMOs and multiplayer games at the time were really that metaverse-like. Fortnite came out. The consoles opened up to cross-platform. Voice chat became a big thing. Squad play became a huge thing. Roblox began its growth too. If you look back over the last six years, you can see the metaverse audience grow from a negligible number to about 600 million. It’s easy to overlook that. A lot of people say Roblox is just a kids’ game or Fortnite is just a shooter. But there’s an emerging market here that will reach billions of users. We should appreciate that and separate it from the unfounded hype that was given to it, the claims around it, NFTs and things.

Sax Persson: If there’s a better term than “metaverse,” sure, we should use it. But there just isn’t. It ends up describing the effect that we think is very beneficial. It’s the multiplayer aspect, the meaningful social play, the meaningful choices and meaningful participation. It’s not owned by these institutions that make you play what they want you to play. I think 40% of Fortnite is now spent in user-generated content. That’s enough for us to pay attention. That happened almost despite itself. These were fun, creative tools, but they’ve turned into a primary activity for a lot of players.

The bet we’re making is to allow creators to feel like part of the platform, to the point where they feel they can bet on it themselves. They can invest in it. We’ll expand the way that we think about meaningful social play between different groups, different ages, different cultures, different areas. You can have your group playing the games you want that aren’t games we would ever make. A story-based RPG, say. That’s not what we’re going to make. Somebody’s definitely going to make it with UEFN, though. We’ve bet there’s going to be an audience that would never come to us for the shooter, but they will come for quality content that they like.

GamesBeat: What tech do you think is ready for metaverse applications today, versus what you still think is maybe farther away? Do you separate the vision in that sense – the metaverse now versus the metaverse later?

Sweeney: You need a lot of building blocks. The metaverse will be a lot more complex and sophisticated than the world wide web, for example. Partly because it has to compete with every other entertainment medium today, including triple-A games and social networks. It can’t fall short of the best games of the day.

I talked about standards in State of Unreal. Right now we have good working standards for content authoring interoperability. A developer can move content from Blender to Fortnite or Unity. All these different 3D packages fit together rather seamlessly. But we don’t yet have technical interoperability at the user level. Fortnite and Roblox don’t download the same assets from the same servers. But that’s within reach. The different standards groups are working on standards. We’re at the point of having that in the next few years. Then you have the ability for somebody to host an experience, perhaps, that could be visited by a browser that’s running Unreal or a browser that’s running Unity or Godot. That’s within reach.

The lack of connectivity between the social aspects is something – it’s funny. Everybody uses WebRTC, this web standard for voice communication, for the foundation of their voice chat systems. But despite it being a standard, nobody has actually connected our systems together. That’s a step that could happen over the next couple of years.

GamesBeat: Like WhatsApp and Telegram.

Sweeney: What if every console social service’s voice chat and every major publisher’s voice chat worked together? What if they had a single friends system that connected together, so that your friends on one platform are your friends on all the others? That’s all achievable. Then you need economic interoperability, which means the ability to buy something in Roblox or Fortnite and use it in another place. There are business challenges there, because then you need revenue sharing between ecosystems based on engagement. You need some sort of governing bodies to carry that out, much like the ratings groups today that rate games for age-appropriateness. You need that for metaverse cosmetics. You need quality standards and safety standards for interoperable metaverse ecosystem destinations. We could have that by the end of this decade. These things could be connected. We could have the open metaverse.

Persson: Companies would get value from the size of the social graph. The social graph is this precious thing you keep to yourself now, but the truth is, the real power would be if these social graphs overlapped. You’d have a much better chance, if we had cross-play, that the people you want to connect with are part of the system you’re in, whether it’s Telegram or Facebook or something else. The idea of having to be distinct members of all these different societies in order to participate just seems antiquated and anti-user in many ways.

SuperAwesome is part of Epic. They have an interesting experience around verified parental consent, where a lot of companies are subscribing to a common database of verified parental consent to sign in discreetly. It’s one step toward–why not implement industry standards for the best way for us to do verified parental consent? What’s the best way to form a squad, the best way to go between these different games and experiences and ecosystems? You just inconvenience your players every time you ask them to sign in for this, sign in for that.

GamesBeat: How do you want to figure out the lines between things that are completely open, things that are more like a de facto standard, and something proprietary? USD is a good 3D asset standard. Omniverse is a way to connect all those things. But that’s proprietary to Nvidia. Does that need to be part of what’s open, as opposed to one company owning this Omniverse interoperability?

Sweeney: The ultimate destination is open standards for everything. The path to that involves a lot of chaos. There’s going to be a lot of de facto standards emerging as different companies propose different things. People will say, “That’s cool. Let’s use that.” That will happen before a standards body steps in. That’s a healthy process. We’re pitching Verse as a potential metaverse standard scripting language. Roblox could pitch Luau, their scripting language. There could be a robust debate around that. Different companies can try different things and we’ll see which ones prevail and become the standard.

GamesBeat: You briefly mentioned Verse. Can you describe it a bit more and why you see it as significant or necessary?

Sweeney: The metaverse needs a programming language, just like the web needed Javascript. The metaverse needs a language that people can use to define how objects behave and interact. In Roblox that’s Luau. In Unity it’s C#. In Unreal it’s historically been C++ and now in Fortnite it’s Verse.

We took a very long look at all the issues facing programming in the metaverse and came to the conclusion that the metaverse would benefit from a language with certain characteristics. Because metaverse means lots of developers deploying content to billions of users, you want a language that can catch problems at development time rather than after release. You want a strongly typed language that has quite a lot of static verification capabilities. The metaverse is going to have a huge audience of programmers, so you want a language that’s easy to learn. Javascript has proven easy to learn. Python has proven easy to learn. You need a language that’s at least that easy, but hopefully even easier.

Two other unique things about the metaverse. One is that time passes during gameplay. You need a language that’s good at dealing with things that occur over a period of time. That’s been a source of challenges in languages like Javascript. Asynchronous programming is hard. One of the major design points of Verse is some great asynchronous programming capabilities. You can say, “Wait for this event to happen and then do this some time later in gameplay.”

The final thing is that the metaverse is not going to be an app store. Metaverse content isn’t going to be an app. An app is built by a developer. They use a bunch of modules in their development. Maybe they use Unreal Engine, even. But ultimately they compile it down into a single program, and then they release it. That program doesn’t change again until they release a new version. The program is entirely under their control.

But the metaverse is going to be a mashup of millions of objects defined by millions of different programmers and content creators. They all need to work together, even in cases where you have a car written by one programmer interacting with a human character that’s written by another programmer. These programmers have never talked to each other. They’re just using some common interfaces. The metaverse needs a massive scale open world programming environment that’s designed for software that’s continually evolving in a real time environment.

GamesBeat: Do you think generative AI is going to be one of these building blocks to get there? It seems like the metaverse needs so much content to be properly considered a metaverse. You don’t want something that’s barren to be a place where people hang out.

Persson: The world needs assets. You saw the opening presentation called Opal, with Rivian driving through. Procedural capabilities are used today for fantastic world-building in general. I don’t know what our position on–is it needed? No. I don’t think so. Is it helpful? Could be. It’s not a great focus of ours, though.

GamesBeat: To me it seems like on the UGC side of things, there are a lot of people who have lots of ideas for games and applications, but they don’t have the skill. The generative AI part is going to help them out.

Persson: I was talking to someone on the show floor who founded a company to make birthday parties with UEFN. It’s essentially procedurally generating an island for Johnny’s birthday party for seven people, with invitations on the table, songs that everyone sings. It’s awesome. Is that generative? No. But do we help? Do we encourage people to make things by making creation easier? Absolutely. Not everyone is an asset developer. We want the smoothest way for entrepreneurs to make the things they want to make. You can only do that if you have an open system. What we announced is just one way to get assets in. It doesn’t stop you from importing whatever assets you want to put into UEFN.

GamesBeat: Would you expect Verse to be down on that user level, though? Is that a user-generated content tool?

Persson: People can make code modules that other people can use and customize. I like to use elevators as an example. One of the most difficult things to make in any game is an elevator, because it has the most complicated mechanisms. Coming and going, opening, closing, not squeezing the player. Getting the camera to work. So many things happen in an elevator in a game. If someone made an elevator for me that I could put another skin on that I could use in my game, that was programmed in Verse exactly as an elevator should function, of course I’ll take that elevator.

Programmers with Verse are going to be asset creators, essentially, amongst many asset creators, making the things that I want when there’s no need for me to make them again myself. The vision is that it’s not just textures and assets. It’s logic too.

GamesBeat: When it comes to solving the really hard problems like the real time sniper in the metaverse – when you have thousands of people in a dense space for a concert, all of them interacting with each other – some of that seems like you need a new internet. We need 10-gig cable across the whole world, something like that.

Persson: Communicating with large-scale concurrency, yeah.

GamesBeat: I wonder how far you think some of that needs to go in terms of fulfilling the dream of the metaverse.

Sweeney: That’s a programming language problem. Right now Fortnite battle royale is limited to 100 players per session. That’s not because 100 is the perfect number for the genre, but because that’s as many players as we can run on a single PC in a data center. Battle royale simulation does not distribute among multiple nodes in the data center. We’re limited to one.

Previously, when game developers wrote simulations like MMOs that scale across many computers, the programming model has been a terrible mess. You have object duplication bugs in MMOs and all these other things going wrong as a result of having a complicated, obscured programming model. I don’t think that’s a possibility for the metaverse. All of our aims here – we’ll talk about this later in the Verse tech talk today – are around our efforts on transactional memory as a future solution to this problem.

The idea is you’d like to have programmers be able to write normal gameplay code, and then have the programming language and the hardware environment cooperate together to scale that simulation across all the nodes in the data center. You make it the runtime system’s problem, running that code concurrently and sorting out its dependencies. We think solving the metaverse problem will involve new programming technologies that just don’t exist today. That’s something we’re working to do. We feel we critically need to do that in order for this to succeed.

GamesBeat: If I were to be a soldier in a gigantic Lord of the Rings battle, would this be a metaverse application? Is that something we should strive to deliver for gamers?

Sweeney: These are just LOD problems. We can have billions of polygon scenes rendered in real time through Nanite. The engine has systems to scale it down to something that looks indistinguishable from the real scene, but simplified. That’s a problem for the network layer as well. But it’s really up to game designers how each experience should work. Some parts of the metaverse will be sharded. If you want to have a concert that’s more like a dance hall, you might want to have 300-player sessions and thousands of them running concurrently, rather than one session with 3 million players. That might be more fun. Designers will have to make those decisions. They’ll vary from place to place and experience to experience. It’s not just forcing everything to be an MMO.

Persson: To the point about whether we need a new internet, I think we’ve concluded that there’s no internet that can scale to this ambition. You just need a different paradigm for how you think about the execution of your program. Verse is conceptualized in the long term to be that. Just going down the road of–you can get better and better. I’ve been in 10,000-person games like what Improbable is doing. That’s really cool, but it can’t go to 100,000. Physically it doesn’t scale anymore. It just needs a different paradigm.

Part of what we’re aiming for is to find solutions to problems that everybody understands where we face the limits of physics. We have to rewrite some more of the stack in order to do this. We can’t just rely on the monolithic structure of the world that we’ve always relied on. It’s just not possible.

GamesBeat: How do you get more people to adopt a new programming language? Some people might react by saying that they don’t want to learn another one.

Persson: We have to prove it’s worth it. I don’t know if there’s any other way for any language to succeed other than showing the value. Write it once and it keeps working. That would be nice. That rarely happens in programming. That’s part of it.

We’re lucky that we have Fortnite. That creator community is getting programming capabilities for the first time. We’re not asking them to switch over right now. We’re asking them to come on board as we build out the language. That gives us a chance to battle test it. It gives us a chance to not wait 10 years before we ship the perfect language. We can adapt as we go along with a very captive audience. They have all the incentive in the world to be a part of this.

GamesBeat: Do you have some hope that the open metaverse will happen? Do you have some reassurance that you’ve gotten from the marketplace that we’re not going to end up in the walled gardens again?

Sweeney: The distribution monopolies, the iOS App Store and the Google Play store, will try to keep their paywalls in. They’ll try to institute roles that force everyone to funnel all metaverse revenue through their increasingly steep paywalls and advertising service monopolies. That has to be stopped in order for the metaverse to have a chance to compete economically. Otherwise they extract all the profit out and there’s nothing left for those who are investing and building the damn thing. That’s a real barrier.

The Metcalfe’s Law argument is dominant here. This isn’t an entirely new market. There are three major console companies. There are two major computer platforms and two major mobile platforms. Then there are dozens of publishers with their own ecosystems and their own player bases. No one company is in a position to build a monopoly, a metaverse monopoly, and force everyone else to use it. It’s going to be companies working together and forging alliances to build either corporate partnerships or an open metaverse together that’s going to happen.

We’re on team open. Every willing company that participates in it will be a part of it. Right now, initially, that’s going to be a lot of small developers, but over time more big companies will join it because they recognize there’s just way more opportunity for them if their player base is connected to everybody else’s, and the players are more likely to be connected to their friends. Metcalfe’s Law is real. It’s a force that’s shaped industries and it’s a force that can be harnessed to build a successful open system, to avoid the pitfalls that left the last couple of decades of platforms stuck in these monopolies.

Persson: There was a point when AOL seemed like an excellent idea.

GamesBeat: Will the network more naturally develop in the direction of freedom?

Persson: I think so, yeah. I think you will see companies that you don’t know today that are embracing –that can lean into what’s already happening. It’s tempting to just keep building the wall higher and higher between these ecosystems. That’s the opportunity for new entrepreneurs and new investment to come in. When we talk about metaverse, we’re attempting to lead by example. The economy shift in Fortnite is an important piece of recognizing that in order for us to be able to interoperate, we need to shift the economy away from how we were doing it. We need to embrace creators and we need to embrace the value they generate on the platform. They need to be part of the profits. It’s as simple as that. The only way we can aim toward where we want to go in the long term is if we’re just a participant in our own platform. We can’t be the dominant big bad Fortnite in the corner. We have to aim to be one amongst many or we can’t grow.

GamesBeat: Were there some advances today that you felt like maybe you want to reinforce here? One that I liked was that when the truck rolled through the water, you could see the mud washed off. You don’t expect that to happen. That was pretty cool. But I don’t know what kind of advances you feel particularly proud of that you showed off today.

Sweeney: There’s so much.

Persson: It’s like saying which one of your children is the prettiest. I think the metahuman stuff is transformative in, again, making content. Allowing democratization of high-quality content. The jungle demo is amazing. The UEFN capabilities are layering on top of a deep foundation from UE. We’re now extending professional level capabilities to all sorts of creators, big and small. It should feel, hopefully, from the presentation, that a lot of these investments that Epic has made are converging on a single goal. The single goal is building a persistent metaverse space where the assets and technology and economy are all accessible to anybody who wants to participate. It should feel seamless over time. It’s just not seamless today.

Sweeney: The convergence is happening now. A lot of the features we show off here are born out of game developers looking at what we had previously and saying, “This isn’t good enough.” But increasingly it’s coming from automakers or filmmakers saying the same thing. When you build something that’s good enough for a feature film, the power that gives game developers, including in Fortnite now, it’s astonishing the amount of power that comes from crossing over among these mediums.

GamesBeat: Hollywood and games is one of our themes for an upcoming conference. One of the people who was helping us with that, Jon Goldman at Skybound, was saying that he sees Hollywood and games between one ecosystem finally. We have The Last of Us on HBO. The ecosystem seems like it’s finally working the way people wanted it to work in the transmedia days. It feels like the point of that — the logical end game, maybe — might be the metaverse. If you can get these different industries to work together as one industry now, you’re checking off movies at the same time you’re checking off PC, console, and mobile. You have different doors into the same worlds.

Persson: You’re describing our vision. A single asset among all different types of media. It’s an important proof point. Does the pipeline scale to any use that anybody would ever have for this asset? That’s a basic requirement for anything we do. It depends on what lens you put on Epic, though. With State of Unreal, it’s the Unreal lens. You see it from the technology perspective. But if you put the consumer lens on it, it’s exactly what you just said.