"If you haven’t put AI in your pipeline yet, the train will pass you by"
Google Cloud’s Jack Buser on how AI can reshape development processes and player experiences
Hello! And a special welcome to our many new subscribers. In AI Gamechangers, you’ll find regular conversations where we spotlight the innovators bringing AI into games. Thanks for joining us.
This week’s is a big one, featuring an exclusive one-on-one conversation with the head of games at Google Cloud. We sat down with Jack Buser, a veteran of PlayStation and Google Stadia, to talk about what Google Cloud is doing for AI and games. From his unique vantage point, he shares the results of a recent poll showing nearly every studio is exploring AI. What does this mean for both development pipelines and player experiences?
But first, two quick announcements:
1/ You can get six of this year’s AI Gamechangers interviews in ebook form, ready to download to your device as a PDF for easy reading on the move. Head here to grab it.
2/ There are just a couple of weeks before the AI Gamechangers Summit takes place live in Helsinki. Don’t miss out: here’s your discount code (AIGAME10N). See you there!
Jack Buser, Google Cloud
To understand where AI in games is going, it helps to talk to someone who has already been at the forefront of major technological shifts. Jack Buser, Global Director for Games at Google Cloud, is that person. With a three-decade career spanning Dolby, PlayStation, and Google Stadia, he’s been a key architect behind game streaming, audio, free-to-play, subscription, and virtual world platforms. Now, he's applying that experience to the next revolution: AI, thanks to Google Cloud's services for game creators.
He brings us data from a survey conducted with The Harris Poll, revealing that 90% of developers are already embracing AI. But Jack looks beyond the tools, explaining how it's transforming development from the ground up. His central focus? Unlocking the "new player experiences" that were once impossible. We sat down with Jack to explore the implications and get answers around issues like security, cost and scale.
Top takeaways from this conversation:
A recent survey found 90% of developers are already using AI in production workflows, far higher than previous industry estimates, with applications spanning from pre-production to QA testing.
The key technical challenges for AI in games are scale, latency, and cost management, requiring sophisticated infrastructure orchestration between on-device processing, cloud deployment, and managed AI services.
AI-native game studios are achieving dramatic efficiency gains using Google's enterprise-grade AI platform Vertex AI.
AI is enabling new game mechanics, smarter NPCs, and even entirely new genres. Indie studios are leading the charge, experimenting with AI in ways larger studios can’t.
AI Gamechangers: Please tell us a little bit about your background.
Jack Buser: I’ve been in the games industry for going on 30 years. I joined in the late ’90s, working on some very early games, including N64 and even PlayStation 1 games.
I got my start through a company called Dolby Laboratories. You’ve seen Dolby Surround sound in the movies; I basically convinced them to get into the games industry, because at that time, games were getting higher fidelity. They were going high definition, ultimately. And I convinced them, saying, “Look, games are a legitimate form of entertainment, every bit as much as movies and TV shows, and so we should bring the same great audio to these games that we brought everywhere else.” I did that for a number of years. We had our technology in all the big consoles, from the original Xbox to PS3, GameCube, you name it.
Fast forward to PlayStation. I spent nearly a decade there running their digital platforms. I ran PlayStation Home for some time. That was a virtual world built into the PlayStation 3. It was actually the first free-to-play platform on a game console. So we had to do a lot of innovation back then to figure out what free-to-play means. It was still a very, very fresh concept in the games industry, and how we brought that to console had never been done before, so that was a lot of fun.
“You’d better believe some of these small companies are going to be the next giant game companies. We’ve seen this through every technological shift in this industry”
Jack Buser
We built the Instant Game collection in PlayStation Plus, which you may remember was probably the first game subscription on a console. Then, I helped build PlayStation Now, which was the first game streaming platform on a game console! I had a pretty broad remit there at PlayStation. Basically, everything digital was likely something that I had touched.
Then I joined Google. I was one of the founders of Stadia, the first industry person they brought in. There were 10 of us at the time. “We’re going to build a game console!” became “It’s going to take more than 10 of us, folks!” So, I worked with Google to show them how to build a game platform for AAA games and how to scale it up. That was a ride of a lifetime, and then ultimately, I was tapped to come to Google Cloud, where I’ve been for a number of years now, to lead our games business globally.
Tell us about that role, please?
I’m the Director for Games at Google Cloud, and my role is to bring the best of Google technology to the games industry. Essentially, to take the same technology that powers our big live services like YouTube and Search and all this other great stuff, and bring that to video games via Google Cloud, making that enterprise-ready and something that an external company can use.
“Mobile developers are in a very good position to leverage their pools of data for AI use cases, not just in development, but also in player experiences. I really see a very bright future for AI on the mobile side”
Jack Buser
The idea is: games are online. They’re connected. They use many of the same services and use cases as our other live services. So, essentially, we give them what we use to create bulletproof online backends. An online game can launch without worrying about it. It’s not going to have to come down for maintenance. It’s not going to fall over at launch or anything like that, and once they get up and running, they have bulletproof analytics. Their game servers can scale to meet any amount of demand. They’ve got databases that can keep track of everything that’s going on in the game.
Most recently, it’s been about AI: AI for game development, AI for new player experiences. We spend a lot of our time now not just helping game companies with core infrastructure powering online live games, but actually helping them leverage AI in their development pipelines.
And ultimately, my favourite part is actually creating new player experiences.
You recently released a report: “New Google Cloud Survey: Gen AI’s Impact on Game Development and Player Experience”. Google Cloud collaborated with The Harris Poll to issue a report that reveals insights into how game developers are embracing AI. What are your highlights?
When we first got that report back, I was surprised. The numbers were much higher than I had expected. We had seen previous reports that had numbers that were healthy, but seemed to sort of understate what we were actually seeing with our customers. When we got those numbers back, we were seeing things like roughly 9 out of 10 developers think that their players’ expectations of what a game can be are changing due to AI. They believe the industry is changing. They believe that their development pipelines are changing. All these data points were hovering around basically 9 out of 10, which is fairly unanimous for anything in this industry.
Even looking at the players, you can see that about a third of them have changing expectations as well. They’re saying they want their games to be more alive, more dynamic. And so as a result, it really underscored the sentiment and the reality that we were seeing on the ground when we work with developers.
If you know about Google Cloud, all of the Big Ten game companies work with us. (If you were to rank them by revenue. They’re all using Google Cloud.) And so we’re talking to all the big game companies day in and day out, as well as the torso and tail of the industry. Without exception, these conversations are about AI. And then this Harris poll came back, and that reinforced what we were seeing in the field.
How are people using AI in their studios to make games?
It really depends on the developer. Our message to developers is pretty simple: if you haven’t started putting AI in your development pipeline, you should probably do that. If you don’t, the train is going to pass you by.
You saw in the survey that the vast majority of developers, roughly 90% of them, had actually put AI into production.
“The cost of development has increased by 90% over the last few years. It’s outpaced consumer spending by 1.6x. That is not a good situation for our industry to be in. What’s out there to help? We are fortunate that AI has met this moment”
Jack Buser
If you rewind a couple of years, the message was, “What is the difference between using consumer-grade AI – the websites that any person can use – and enterprise AI?” And, “How does enterprise AI help protect my data and ensure that things are safe and secure to use?”
We started educating developers a couple of years ago. Yes, there is enterprise-grade AI, and it’s not just used by game companies, but by banks, financial institutions, and governments. This stuff is bulletproof and built for enterprise. All of a sudden, you started to see a willingness to build proof of concepts within the development pipeline. Maybe not to put it into full production back then, but to actually start to train models on their data and try to figure out where this might fit in the development pipeline.
Fast forward two years, and we’re seeing every single use case between pre-production, full production, and QA impacted by AI. We have what we call “AI-native” game companies that we work with. These are game companies that have basically been founded in the last couple of years, and they’re founded on this notion that they’re going to leverage AI in everything they do, from game development to player experiences.
One of them, Series Entertainment, put out a press release not long ago saying they had taken our enterprise-grade managed platform for AI (called Vertex AI), and tried to use that across their entire development pipeline. They reduced their development cycle by 90%. Time equals cost, right? Now, that’s very extreme! That’s obviously a very extreme example from a company that’s very AI native.
But even if you look at very large and sometimes very conservative development studios, they absolutely have put AI into their production pipeline, if only for pre-production, just getting ideas out there. But increasingly we’re seeing it used for full production; everything from asset generation, dialogue generation, and then even quality assurance, play testing, and balancing.
And you saw that in the data. Developers were coming back saying, “AI is great for play testing, AI is great for localisation, AI is great for code generation.” It’s what I talk to my team about all the time. You can build out the entire development pipeline. Obviously, each one’s different, but there is not a single use case that’s not going to get impacted by AI or hasn’t already, frankly.
Studios will be conscious of security and privacy. Developers often want to secure their code and run it on premises, rather than upload it to an AI model somewhere online. What’s Google Cloud’s answer to that?
In the old days, game companies were very worried about using the cloud at all. They wanted to lock their data down. “We’re going to rack our own servers in a building. We’re going to lease a building and build walls around a physical thing.”
Over the last ten years, they’ve started to realise they’re not in the data centre business, they’re in the game development business. They started to realise that, cloud is enterprise-grade. Yes, you are using computers that are owned by Google, but they’re walling this stuff off. It is running in your own Google Cloud tenant, and it’s, frankly, the most secure you could get. This is absolutely your data; no one else can see it. We can’t see it! It is completely walled off and safe. That happened even before AI. When you think about your analytics stack, your player information, or your databases, all that is super secure and protected.
“Think about games from 30 or even 40 years ago. Graphics are a lot better, the games are more sophisticated, but it’s still the same old NPC tech we’ve been using for a long, long time”
Jack Buser
When AI came, people had very similar concerns. It’s their data. “We want to make sure that our data stays secure. When we tune these models, we don’t want that data to go back up and retrain the master model!” At the beginning of generative AI, just a couple of short years ago, a lot of development studios were going onto the public web and using AI websites, and their legal departments were saying, “You’re leaking our data out there. Stop it immediately!” That’s when we came in and showed them platforms like Vertex AI, which absolutely protects your data. It ensures that your data isn’t going back up to train the master model. It keeps everything rock solid and secure.
It’s generally eye-opening to a lot of game companies that Vertex AI was not built for games. This was built for industries that are much larger and have security concerns that are much more severe than the games industry. It’s bulletproof. If it can meet the needs of a financial institution, rest assured, it can meet the needs of a development studio.
Let’s talk about AI and player experience. How do you think AI is going to directly impact that?
I’ve been in the games industry long enough to know it’s very, very hard to predict this stuff. What we know at Google is that we have provided technology, tools, and platforms that will enable game developers to create stuff that we can’t even possibly imagine.
We’re starting to scratch the surface of what’s possible out there. You’ll see games that have smart NPCs. These are NPCs powered by AI that don’t follow a simple dialogue tree. You think about games from 30 or even 40 years ago. They’re still NPCs powered by the same old dialogue tree giving you two choices… that hasn’t changed! Graphics are a lot better, the games are more sophisticated, but it’s still the same old NPC tech we’ve been using for a long, long time.
Enter AI, and it’s revolutionising what a computer-controlled character can be. These can either be an AI “friend” of yours that’s sitting on a “virtual couch” next to you and helping you through the game. Or it can be a character in the game that doesn’t break the lore of the game, that knows that you’re stuck, or you’re in a high churn point in the game, and without breaking the world, actually eases you through that moment (without you even knowing it’s AI).
You see examples of add-on content, where you go into the shop and see a bunch of stuff that was made by the development team, and you think, “Well, I like that, but I want it to have a different texture”. And you can just describe what texture you want to have, and it will do it. Or AI could create the entire item.
There’s user-generated content: sandbox games, where instead of using low-code or no-code tools, you can actually use natural language. Just say, “Build me a world that looks like this”. And it will just create it.
But again, I have to stress this is just scratching the surface! These are new game mechanics that are being applied to traditional game types.
“Vertex AI was not built for games. This was built for industries that have security concerns that are much more severe than the games industry. If it can meet the needs of a financial institution, rest assured, it can meet the needs of a development studio”
Jack Buser
Because game development takes years and years, we’re working with game development teams that have games that won’t see the light of day for years to come. What we’re seeing is a new class of games where they’re creating entirely new genres. Even before they hit pre-production, they’re saying this game is going to be AI native. “We’re going to build this game design from its core to take advantage of new play mechanics.”
You’ll see basically three things. You’ll see new mechanics added to traditional games. You will see entirely new game designs that don’t fit existing genres, and you’ll see stuff in between, where it’s a new take on an existing genre, but it’s radically changing what that genre could do. And we’re seeing everything in between. It is very hard to predict where this all ends. It’ll be game development studios that take us there: the creative geniuses that are actually making the games. But rest assured, it’s coming.
There’s a cost to using AI. How does a studio keep control of the cost and manage scale?
The very first concern that game studios often have is around security. We explain Vertex AI and get them over the hump, then three more issues come into play, where we spend the vast majority of our time with game companies: scale, latency and cost.
And again, we’re in the context of new player experience. It’s not necessarily game development, because game development is actually on a much smaller scale, with much smaller cost, and latency requirements are much less stringent. But when you’re talking about new player experiences, you want [the AI] to respond, and you can have millions of players. You can imagine huge spikes.
“There’s user-generated content: sandbox games, where instead of using low-code or no-code tools, you can actually use natural language”
Jack Buser
Essentially, what we do is we work with game companies to first define the experience they want to build, and then help them architect the proper back-end technology to allow them to meet their cost, scalability and latency goals. It gets very technical, but we’re able to orchestrate where this inference actually happens and how that inference is actually being handled, in order to control those three parameters.
On one extreme, you could deploy the inference on-device, right? And that’s great in many ways, because the consumers have already paid for the device. Cost can plummet when you do that. However, sometimes performance will suffer. You might actually not have an authoritative experience. If you have the inference happening on your device, but you’re playing with several people, somehow the result of the inference on your device has to get to all players instantaneously, and so the latency of doing that might be prohibitive.
The other issue is that the resources on the device might be challenged. You need a pretty beefy GPU to perform any amount of sophisticated inference in the first place. Even if your device has the power to do it, that GPU is also rendering the graphics, and on many devices, there’s a battery involved! So again, there are a lot of considerations.
(When I say inference in all this, I mean it’s that last stage of AI where it’s generating and returning the response.)
Now, go to the other extreme, which is the cloud. Vertex AI is our managed platform for game development. It has the ability to deploy these inference endpoints, and it’s very automated. You don’t have to think too much about infrastructure or anything like that. All you need to know is that you can say, “Vertex, deploy this endpoint,” and it essentially takes care of it for you, deploying this inference endpoint in the cloud. It has all of the great benefits of the cloud. It’s authoritative, it’s low latency, it scales to meet any number of players; but you might not have direct control over the actual infrastructure beneath that.
“You might have development studios that just never use AI, and that’s okay! They all have their own reasons. I don’t expect this to necessarily get to 100%. I don’t think that’s the goal”
Jack Buser
Now, you know game developers. If there’s one thing the big game developers are, it’s highly technical. We find a number of game developers want to meet somewhere in the middle. Maybe they want to manage the inference themselves, or the infrastructure underlying it; they want to be able to control it. They use a technology called Kubernetes. We have our own flavour of it, Google Kubernetes Engine (GKE), and essentially, what they’re doing is they’re treating inference like a game server. It has a lot of the same computer science that’s being used to handle these huge game server simulations in the cloud: low latency, scalable, cost-effective. It’s basically the same computer science problem for infrastructure, so they’re able to handle inference much like a game server, where they’re managing the infrastructure underneath, and then you can orchestrate across the three.
You can say, “For this experience, I want it on-device.” Or, “For that experience, I want it in the cloud. I want Vertex to deploy for this experience.” Or, “I need to manage the infrastructure because of reasons.” It gets pretty sophisticated. This is what we spend the vast majority of our time on currently, making sure we’re able to do this cost-effectively and at scale.
Google has a leading product in the AI space, with Gemini and its related products. What if a developer is committed to a different frontier model? Can that also be deployed?
This is the beauty of it. Vertex AI has Gemini in it. It’s the enterprise-grade version of Gemini. It basically has all the parameters of Vertex AI that I was speaking of in terms of security and enterprise-grade, but it doesn’t just have Gemini; of course, it has all the foundation models from Google inside of this platform. Then we have other models from other third parties, and then we have open source models in there!
It doesn’t matter what model you want to use, Vertex AI is a very robust platform that has what we call the “model garden” in it, and you can choose exactly what model you want for any particular use case. What’s cool for a lot of game developers is that it’s also connected to Hugging Face. If you explore around Hugging Face, you can find so many cool, open-source models, and again, you get all the benefits of Vertex with that connection as well.
Going back to the survey, what about the 10% of developers who aren’t convinced about AI yet? How do they factor into your worldview, and how do you bring them around?
It’s like so many technologies that come into the entertainment industry, or any industry whatsoever. You might have development studios that just never use AI, and that’s okay! They all have their own reasons why not. I don’t ever expect this to necessarily get to 100%. I don’t think that’s the goal.
I think the challenge we’re trying to solve right now is twofold. Our industry has returned to health to a certain degree. Those years coming out of the pandemic were a surprise; I’d never seen this industry actually contract. This is the first time, I think, probably at least in the US, since 1983. But now we’ve returned to a certain amount of health; $188 billion last year, 2% give or take, year on year, growth. That’s not double-digit growth or anything like that, but that’s growth. We’ll take it. It’s not contraction anymore, but that’s being met with a couple of challenges.
“The mobile industry has tended to be at the forefront of many of the biggest trends that we’ve seen in the entire industry over the last five to ten years”
Jack Buser
The first, extremely severe challenge, is that the vast majority of play time is actually in older games. If you’re creating a new game, you'd better do something new, otherwise, you’re not going to attract people out of the small number of games that have been around for six or more years. If you’re operating one of those games… well, you better do something new too, because if you don’t, how do you keep them on that platform?!
Challenge number two is probably even more serious, which is that the cost of development has increased by 90% over the last few years. It’s outpaced consumer spending by 1.6x. That is not a good situation for our industry to be in.
What’s out there to help mitigate these two big challenges? We are fortunate that the state of the art in AI has met this moment. We have tools as an industry to help. I think had AI not lined up with these challenges, we’d be in a very different spot right now as an industry.
Does everybody have to use AI? No, absolutely not. But most studios, the vast majority of our industry, are struggling with these challenges right now. It tends to make us very mission-driven with this stuff, because we all love games. We want to do what we can to help.
One of the areas which has bounced back better than others is mobile. How do you see AI, and Google Cloud particularly, supporting mobile development?
Mobile is such an interesting story, because obviously we’re a huge player on mobile, not just in Google Cloud, but also in Play Store and with Android! The mobile industry has tended to be at the forefront of many of the biggest trends that we’ve seen in the entire industry over the last five to ten years.
Start with live service games. Free-to-play. Sophisticated analytics. User acquisition, retention, monetisation loops. All of these things came from the mobile industry and were later adopted by AAA and other forms of games. Mobile has always been at the forefront of pushing how we got our start as a cloud platform.
Platforms like BigQuery, that’s our platform for game analytics. It’s got AI built in so that you can better understand your players and your business. It was really mobile in the early days! We needed extremely sophisticated analytics tools because our game designs and the way we operate our live service businesses are based on our understanding of our players. We’ve seen mobile developers push our analytics tools further than any other category of developers, by a lot.
“We’re talking to all the big game companies day in and day out, as well as the torso and tail of the industry. Without exception, these conversations are about AI”
Jack Buser
As we enter AI, mobile developers are actually in a very good position to leverage their pools of data for AI use cases, not just in development, but also in player experiences, because they tend to be very sophisticated with that kind of stuff. They’re starting from a position of advantage, whereas you can still find some AAA developers who are trying to figure out how to best use their analytics stack. They have all this data, they don’t quite know what to do with it. But I think by and large, for especially the more sophisticated, more successful mobile developers, analytics is at the core of their game designs. I really see a very bright future for AI, specifically on the mobile side.
We’ve talked about the biggest studios that might have enterprise-level needs, but what about indies? They’re the lifeblood of the industry. How does Google support the smallest teams?
We’ve thought a lot about this. For many years at Google Cloud, our technology was targeted at the largest game companies in the world. We were taking technology that was powering seven live platforms with over two billion users each, and we were making it available through Google Cloud to the games industry. That was our strategy, and these technologies were very well suited to big game companies that had massive track traffic spikes and huge security needs.
Enter AI, and all of a sudden, we started to see innovation happen with studios that were very small at a much faster rate than we were seeing in the big studios. We started to see these AI-native startup game companies, sometimes less than ten people, create new player experiences that were far beyond anything else we were seeing. And we realised, “Gosh, there’s something fundamental happening here!”
While at the time we were surprised, in retrospect, if you look at shifts in technology in the industry, it’s not unusual. Small, independent developers are going to be the companies that push the innovation bar faster than anybody else. They’re going to be the ones that put AI into production much, much faster, because they can. They’re much smaller and more nimble in terms of new player experiences. Oftentimes, these teams have the freedom to experiment with creative ideas that might be prohibitive if you’re creating, for instance, an annual franchise, where you can’t really take those kinds of risks.
“Games are online. They’re connected. They use many of the same services as our other live services. So we give them what we use to create bulletproof online backends”
Jack Buser
We’ve become very involved with not just the torso of the industry, but also the long tail of the industry, hunting for these game companies that have this vision and this passion for creating these pipelines, creating these innovations, and we’re elevating these companies. We’re essentially putting these small AI-native game companies on a pedestal to show the industry what’s possible.
I will make a prediction: you’d better believe some of these small companies are going to be the next giant game companies. We’ve seen this through every technological shift in this industry. Think about the shift from cartridge to CD-ROM. You had some companies that were unheard of, that were born in that era, and now are some of the largest game companies in the world that were “CD-ROM native”. Some game companies couldn’t make that switch; they couldn’t move fast enough to change their development pipelines, change their player experiences, to take advantage of this new storage media. Take the shift from 2D to 3D, take the shift from offline to online. The story keeps repeating itself in the games industry, and I think AI is going to be no different.
Further down the rabbit hole
Some useful news, views and links to keep you going until next time:
AI-powered Pengu app maker Born raises $15 million. The funding will support a new US office in New York and a new social AI product.
Anthropic will pay out $1.5 billion to authors in a settlement over copyrighted data used in model training. The preliminary approval hearing is set for Thursday, 25th September. The Authors’ Guild has summarised what it all means.
LocusX raises $2.1m for AI-powered QA and bug-fixing tools.
Conversations with leaders from Brimstone AI, NAK3D, AI Guys, Wishroll, Saga and Code Maestro are featured in the free AI Gamechangers ebook 2.
Animoca Brands has released a report on ARAI, an AI system designed to act as a co-pilot in blockchain games, starting with their web3 strategy game Astra Game 2, where it will handle resource optimisation and decentralised financial operations while learning player preferences.
PC Gamer responds to the discovery that top-down satirical shooting game Doomscroll was vibe-coded.
Google’s parent company has announced a $6.8 billion investment in the UK for AI. The money will be used for infrastructure and research over the next two years, with a new data centre opening in Hertfordshire this week.
The International Game Developers Association’s Interactive Audio Special Interest Group launched a new Artificial Intelligence Working Group this week. The programme will have a particular focus on best practice regarding AI audio.
Meanwhile, the French actor who provides the voice of Lara Croft has issued a formal notice to Aspyr, the producer of Tomb Raider 4-6 Remastered. It alleges the games reproduce her voice, using AI, without her consent.