When Trevor Harries-Jones joined OTOY as COO five years ago, what drew him to the company wasn’t just the technology – it was founder Jules Urbach’s vision of a future where computing power could be democratized through decentralization.
That vision, manifested through the Render Network, is now processing 1.5 million frames per month for clients ranging from Hollywood productions to NASA, with 63 million frames rendered to date on a network of thousands of consumer GPUs scattered across the globe.
The Problem: Enterprise GPUs are too expensive
While working on a project for Madison Square Garden, Urbach calculated that rendering the content for their big screen would require six months of Amazon’s West Coast GPU compute – but the project deadline was three months away.
“This is just not going to happen,” Harries-Jones recalls Urbach saying at the time.
The fundamental issue was NVIDIA’s business model. Enterprise-grade GPUs were prohibitively expensive for rendering at scale, making traditional cloud computing economically unfeasible for cinema-grade production work.
Urbach dusted off a patent he held for distributed home computing, and Render Network was born in 2017.
Render ELI5? Torrenting, but for GPUs
At its core, Render Network operates on a simple but powerful premise: send each frame of an animation to a different computer around the world, completing in hours what would take weeks on a single machine.
“So it’s kind of like an initial torrenting idea,” I say. And Harries-Jones agrees. “We stream different frames to different nodes around the world.”
The network taps into our magic internet money economy – idle consumer GPUs that would otherwise sit unused. Anyone with a performance-grade graphics card and a Windows computer can contribute to the network and earn passive income.
This isn’t just theoretical. The Sphere in Las Vegas, that massive LED venue that’s become a Vegas landmark, relies on Render Network’s distributed infrastructure to process content that couldn’t be rendered through traditional means.
Real users, real revenue
In an era where crypto projects often struggle to demonstrate utility beyond speculation, Render Network stands out for generating hardcore revenue from actual users.
“Real users,” Harries-Jones emphasizes. “To me, what I get real value from is what’s created every day. If you go by a booth and see the quality of these creations, it’s phenomenal.”
The network serves thousands of motion graphics artists working on Hollywood productions, concerts, games, and large-scale displays. Unlike GPU marketplaces that compete for scarce high-end hardware like H100s and B200s, Render focuses specifically on serving motion graphics artists with cinema-grade productivity needs.
“We are not a GPU marketplace, per se,” Harries-Jones clarifies. “We’re really a service for motion graphics artists.”
The AI convergence: 2026’s big shift
While rendering remains the core business, Harries-Jones sees 2026 as a watershed year where AI finally becomes genuinely useful for cinematic content creation.
Until recently, AI-generated content suffered from obvious flaws – six fingers on hands, lack of continuity between scenes, and an inability to produce hyper-realistic results suitable for professional work. “Hyper-realism is still hard,” Harries-Jones notes.
But two technological developments are changing that equation.
First, a new file format called Gaussian Splats allows artists to save render outputs as three-dimensional representations of objects rather than flat images. These “splats” can be pulled directly into traditional 3D workflows, dramatically speeding up scene creation.
Second, the emergence of “world models” – AI systems like Apple’s Genie 3 that generate explorable scenes rather than static images – creates a natural bridge between traditional animation workflows and AI generation.
“We believe 26 will be a convergence between traditional workflows and AI to the point where it actually becomes usable,” Harries-Jones predicts.
To capitalize on this shift, Render Network has launched a second subnet specifically for AI jobs, running through Docker containers alongside the original rendering network.
Racing to the bottom? Not if you focus on idle compute
As more decentralized GPU networks enter the market – Akash, Spheron, and others – a natural question emerges: won’t this create a race to zero in GPU pricing?
But rather than competing for scarce, expensive training-grade GPUs, Render Network focuses exclusively on idle consumer compute.
“There is an absolute glut of idle compute, and our goal is to get that utilized as opposed to compete for H100s and B200s,” he explains.
This approach sidesteps the capital cost problem. The marginal cost is essentially electricity. As prices decline, Harries-Jones expects volume to increase proportionally, maintaining total network revenue. And since all revenue is burned (in the tokenomics model), the network remains economically stable whether it’s processing high volume at low margins or lower volume at higher margins.
“For us, that’s really the focus – it’s usage,” Harries-Jones says.
The Power problem is actually an opportunity
While data centers face dual bottlenecks – both compute scarcity and power constraints – decentralized networks like Render sidestep both issues.
“There’s not only a compute bottleneck, but a power bottleneck, both of which are solved by decentralization,” Harries-Jones notes.
The network operates on what he calls “moonlighting” – utilizing nodes during their idle time. When centralized nodes are completely utilized during peak hours, users can tap into decentralized nodes across different time zones globally.
Harries-Jones doesn’t see this as replacement infrastructure but as supplementary capacity. With an estimated 40% of consumer GPUs sitting underutilized at any given moment, the opportunity isn’t to compete with centralized computing but to extract value from wasted capacity.
“We’ve got a shortage of compute that isn’t going to change. It’s only going to grow,” he emphasizes.
A message to the community
For Render Network’s token holders and ecosystem participants, Harries-Jones’s message is simple but urgent: contribute and vote.
“We’re a decentralized ecosystem. Everyone, please contribute. Please vote,” he stresses. “It’s really, really important for the project as a whole.”
And for the artists using the platform daily? “Thank you so much. I love your creations. Please continue. Please make that groundbreaking work that really drives our happiness and drives the value of the network for everyone as a whole.”
As AI tools become genuinely cinematic in 2026 and the global GPU shortage intensifies, Render Network’s bet on idle consumer compute may prove prescient.
Shashank is the founder of web3 content studio yMedia and a contributor at Forbes and Open Magazine. He has been involved in the web3 industry in 2013.