Unlocking Decentralized Intelligent Infrastructure for the Next Era of Compute

AI is reshaping the world—but centralized infrastructure can’t keep up. At DePIN Day Denver, five ecosystem leaders came together to explore how decentralized physical infrastructure (DePIN) can unlock new compute, data, and orchestration paradigms. This panel featured insights from:

Evgeny Ponomarev (Fluence)
Sebastian Pfeiffer (Impossible Cloud Network)
Jeffrey Amico (Gensyn)
Stefaan Vervaet (Akave)
Jarrod Barnes (NEAR Protocol)

They discussed how DePIN intersects with AI infrastructure, why it matters now, and what innovations are already emerging across storage, compute, data pipelines, and agent communication.

The Bottleneck: AI Demand Outpacing Centralized Supply

Traditional cloud providers like AWS are investing billions in infrastructure but still struggle to meet exploding AI workloads. Meanwhile, AI needs not only raw compute, but diverse, high-quality datasets, better orchestration, and local intelligence. Centralized systems are rigid, expensive, and increasingly opaque.

DePIN offers a model for crowdsourced, modular, and programmable infrastructure. Just like Airbnb unlocked dormant real estate, DePIN protocols can unlock underutilized GPUs, idle storage, and local sensors fueling the next generation of intelligent systems.

DePIN’s Edge: A Crowdsourced, Crypto-Incentivized Alternative

Evgeny Ponomarev (Fluence) describes DePIN as the sharing economy 2.0: a decentralized market for compute, data, and services. From compute nodes to data labeling to validation, contributors can be rewarded with tokens for powering intelligent systems.

This unlocks three key advantages:

Access to long-tail hardware and bandwidth
Diverse data sourcing for better model training
Programmable value flows to contributors

From consumer GPUs to drone imagery to edge sensor data—DePIN turns real-world resources into digital infrastructure.

Storage & Data: Akave on Onboarding Enterprises into DePIN

Stefaan Vervaet (Akave) emphasizes that the data problem is massive: rich, proprietary datasets often sit unused behind firewalls. DePIN enables enterprises to offload storage to decentralized networks at lower cost, then selectively share data via tokenized access control and S3-compatible tools.

This isn’t just about cost savings. It enables:

End-to-end data provenance
Composability with smart contracts
New data marketplaces and monetization models

Heterogeneous Compute: NEAR’s Developer-Centric Perspective

Jarrod Barnes from NEAR highlights two core shifts:

Consumer hardware is becoming AI-ready (e.g. Apple M-series chips)
Developers increasingly need flexible compute orchestration

Rather than relying on static AWS instances, DePIN allows workloads to be dynamically distributed across a mesh of devices. It enables elastic compute that adapts to the job—and unlocks access for developers outside big tech.

New AI Architectures: What Gensyn Is Unlocking

Jeffrey Amico explains that DePIN enables training and inference architectures that weren’t previously possible. While dense transformer models demand high-frequency node communication, new “mixture of experts” models can run locally across distributed hardware and only sync periodically.

This style of model is:

More bandwidth efficient
Aligned with DePIN’s distributed topology
A potential unlock for global-scale intelligence networks

Technical Challenges: Bandwidth, Privacy, and Interoperability

Still, DePIN infrastructure faces hurdles:

Latency-sensitive training (e.g. for dense LLMs)
Data privacy and secure execution environments (TEEs)
Interoperability with legacy tools

But as Stefaan notes, challenges create design pressure. Enterprises want data access, transparency, and programmable access rights. DePIN enables this with composable smart contracts and blockchain-native access controls.