This panel was recorded at DePIN Day Brussels — part of the global conference series dedicated to Decentralized Physical Infrastructure Networks.
Hosted by Fluence, DePIN Day brings together founders, developers, investors and ecosystem leaders to explore how crypto-incentivized infrastructure is transforming the real world.
Watch to discover the latest innovations, real-world use cases, and emerging trends shaping the future of decentralized infrastructure.
As artificial intelligence becomes more powerful and ubiquitous, the infrastructure underpinning its development and deployment must evolve. At DePIN Day Brussels, pioneers of decentralized infrastructure — Gensyn, Akash, and Lit Protocol — joined moderator Lukas Bresser (Ramo) to explore why decentralizing AI is both a societal necessity and an opportunity to unlock massive innovation.
Why Decentralize AI?
The panel kicked off with a powerful question: Why must AI be decentralized?
David Sneider (Lit Protocol) pointed to the societal risks of centralized AI, comparing today’s situation to the Web2 era. Platforms built on closed, incentivized systems have already fragmented truth and trust. If AI develops under the same principles, the consequences could be dire.
In contrast, Anil Murty (Akash) offered an optimistic angle, likening AI to the PC, internet, and mobile revolutions. These breakthroughs empowered individuals—but only because they were accessible. Decentralizing AI ensures broad participation, fueling open innovation rather than gatekeeping it.
Harry Grieve (Gensyn) broke it down into three drivers: price, scale, and progress. Decentralization removes costly intermediaries, taps into global compute resources beyond data centers, and opens the door to new economic models where AI agents act autonomously and generate value for other machines.
Breaking the Cost Barrier
AI compute costs are notoriously high, especially for startups. Gensyn and Akash are tackling this challenge head-on.
Gensyn enables ML engineers to split training jobs across untrusted, heterogeneous hardware. The system verifies results and recombines them — offering performance and scale without centralized infrastructure. Their focus? Low price, seamless UX, and wide model compatibility.
Akash, described as the world’s first “supercloud,” provides GPUs and storage at a fraction of the cost of AWS or GCP. An H100 GPU on Akash costs ~$1.50/hour on-demand — far below traditional cloud offerings. This pricing unlocks access for AI startups and non-crypto builders alike.
Verification: The Decentralization Tax
But decentralization comes with trade-offs. As Grieve explained, there’s a “decentralization tax”: additional effort and cost to verify compute results.
Gensyn tackles this with a hybrid model — deterministic execution verified probabilistically. They avoid full replication (i.e., 100 devices doing the same job) and instead sample work for accuracy. This approach maintains scalability while keeping costs in check.
Composable Innovation with Web3
David Sneider highlighted another powerful benefit: composability. Lit Protocol enables decentralized key management and access control. Together with infrastructure providers like Gensyn and Akash, it allows new primitives to emerge — like community-owned AIs trained on private but shared datasets.
For example, imagine a subreddit spun into an avatar that aggregates the collective intelligence of 100,000 users. In this model, contributors could get paid when the AI is used — creating a new economy of user-owned, interoperable AI.
Real Use Cases Today
Akash is already powering real AI startups:
- Brev.dev offers a more powerful Google Colab alternative, integrating Ash to access top-tier GPUs without vendor lock-in.
- Venice.ai is building a decentralized ChatGPT alternative on Akash.
- Prime Intellect is pioneering heterogeneous training, enabling models to train across different types of GPUs— a game changer for open-source AI.
On Hype, Dead Ends, and What Comes Next
The panel ended with a candid discussion about AI hype. Anil cautioned against over-indexing on the GPU shortage. It won’t last forever. Instead, the real value lies in onboarding non-crypto users and building intuitive infrastructure.
By focusing on usability (e.g., removing wallet requirements), Akash is creating services valuable even beyond AI — for hosting websites, apps, and more.
The takeaway? AI may be the current driver, but the long-term opportunity is broader: a new decentralized internet stack.
