GDC 2025

Last week, I attended GDC 2025 on my own dime—as an indie game dev wannabe—for the first time in 10 years. My last GDC was in 2015, where I happily met one of my best GDC sessions: Christy Marx’s “From Jem to CastleVille.”

These days, I work in the online advertising industry, where topics like Programmatic IO are more directly tied to what I do. But I spent two decades in the gaming world, and GDC was where it all started for me. My first GDC was back in 2002, which also marked my first time setting foot on U.S. soil. I still remember the days when John Carmack’s Porsche was the dream benchmark for game developers around me.

With Apple no longer holding in-person WWDCs, I decided to return to my roots. GDC felt like the perfect opportunity to reconnect—not as part of a company or on a sponsored trip, but personally, fully self-funded. It was my one-week pilgrimage to the industry that shaped my early career.

It felt… a bit desolate. Not just because I didn’t have anyone to reconnect with after ten years away but because there was a noticeable sense of disconnect in the air. The conference felt like a gathering of industry survivors—indie developers and mid-sized (AA) studios—sustained by funding from a handful of major sponsors. Yet, those two groups didn’t seem to mix much.

In the West Hall, there were plenty of sessions focused on development and craft, but I didn’t spot many people from the “money” side of the industry. I assumed most of them were tucked away in the business meeting rooms over in the South Hall. The divide between creators and funders felt more visible than ever.

Back in the day, GDC felt like a true festival—a vibrant mix of industry celebrities, developers from studios of all sizes, and investors eagerly scouting for the next big hit. Everyone, from engineers to execs, shared the same space, feeding off each other’s energy.

Maybe it’s just the perspective of someone no longer on the inside, but this time, that's what I felt. I saw only a handful of sessions from AAA studios and even fewer deeply technical talks. It made me wonder if this shift reflects how game development has changed—now more akin to filmmaking, with fewer novel technical challenges and more streamlined pipelines powered by just a few dominant game engines.

Do they no longer want to share their breakthroughs? Or is there simply nothing new to show? That was the most disappointing part for me.

Still, as always, there were silver linings. I came across some fantastic indie games, met a few developers pushing the boundaries of what’s technically possible, and sat in on some thought-provoking discussions about the role of AI in gaming. Here are a few highlights that stood out to me.

AI was undeniably a central theme at this year’s conference, and my feelings about it were all over the map—ranging from anxiety over potential job loss to curiosity, passion, a bit of skepticism, and even some genuine appreciation. Traditionally, Game AI has focused on systems that simulate intelligence—like A* pathfinding—to create the illusion of smart gameplay. But this year, the conversation shifted decisively toward Generative AI.

One of the most noticeable trends was the growing acceptance of Gen AI as a tool. The session “Next-Gen Tech Forum: Generative AI’s Honeymoon Is Over: What’s Working for Developers Today?” (presented by Invoke) was particularly insightful. I was surprised to learn that there are already Generative AI artworks considered “copyrighted,” with legal recognition hinging on the degree of human creativity and effort involved in producing the final output. It sparked a meaningful discussion about authorship, ownership, and where the real value lies in the creative process.

Intellectual Property is only developed when there is substantiated human creativity. With AI, an artist or creative must use AI in an iterative process that involves their creativity.

Of course, there were counterpoints as well. One particular session was “The Human Cost of Generative AI,” where Rez Graham tackled the complex balance between innovation and ethics. While he acknowledged the potential for AI to improve creativity and unlock new possibilities, he also emphasized the urgent need for responsible use and thoughtful guardrails. His frustration was palpable—especially when he referenced a signboard casually promoting the idea of replacing humans with AI. It was a stark reminder that the conversation around AI isn’t just technical or creative—it’s deeply human.

Another noticeable trend was the use of Generative AI for in-game content. Inworld AI appears to be one of the frontrunners in this space, though I found their approach to be a bit too ambitious for the current moment. Their vision centers on populating games with real-time, AI-generated content—like dynamically generated NPC dialogue—aiming to create immersive, ever-evolving virtual worlds reminiscent of Westworld or Sword Art Online.

It’s a bold vision, no doubt, but I couldn’t help feeling it’s a bit premature. Most game developers still want a high level of control over the narrative and player experience, and relying too heavily on AI-generated content introduces a lot of unpredictability. Beyond that, it should also be incredibly expensive. Running real-time generative models at scale isn’t just technically complex—it can be financially unsustainable for many studios, especially indie or mid-tier developers. While the tech is impressive, the practical hurdles would remain high.

There were also some refreshingly practical applications of Generative AI. Ada Eden's "1001 Nights" was a great example—an imaginative and well-executed use of technology. That said, even their (not-so-big) project faced significant cost challenges, and it was only made possible through direct support from an LLM provider.

My personal favorite session of this year’s GDC was “Bringing AI NPCs to Life with NVIDIA ACE: On-Device Small Language Models in Meaning Machine’s Dead Meat.” The studio, Meaning Machine, showcased how they built a generative AI-powered narrative mystery thriller. What stood out was their transition from a fully cloud-based LLM solution to an on-device LLM.

Interestingly, they discovered that issues like repetitive dialogue and meandering conversations weren’t exclusive to the limitations of on-device models—these problems were also present in cloud-based LLMs. Their breakthrough came by incorporating handcrafted “director’s notes” to guide the AI’s behavior, resulting in gameplay that felt far more intentional and compelling, even while running entirely on-device.

Thomas Keane, co-founder of Meaning Machine, left a strong impression—and his insights continue to linger in my mind.

(edited) "AI is magic but not a God-like magic, but a magic like a magician’s trick where the effort of that trick is directly proportional to the human effort behind the scene.”

Although I’ve expressed some frustration about the lack of deeply technical sessions at this year’s GDC, “Flex and Fun: Graphics Magic in ASTRO BOT” was a standout hidden gem—and easily my personal runner-up. One line from the session genuinely struck a chord with me:

“Any frame rate drop below 60FPS is considered a bug and must be fixed.”

That mindset alone says so much about their commitment to quality. Tai Yamaguchi went on to share a detailed breakdown of how the team tackled complex visual elements like fluid particles, water surfaces, and mesh deformations. It felt like a masterclass straight out of SIGGRAPH, packed with the kind of deep technical insights I’ve missed at GDC this year.

We need more sessions like this. There was a time when studios everywhere were pushing the boundaries of technology to enhance core gameplay—not just polish. That spirit, that grit, seems harder to find now. It’s time the industry reconnects with that relentless drive for innovation.

All in all, it was a great week. While the first day left me a bit underwhelmed, the conference gradually reignited something deeper—a renewed passion not just for games but for creativity, craft, and curiosity across the board. GDC reminded me why I fell in love with this industry in the first place.

I’m already looking forward to attending again next year. And this time, I want to go beyond just listening—I want to prepare something of my own to share.