In this week’s AMA episode, the host Taariq Lewis, @LewisTaariq, CEO of VolumeFi (volume.finance) is joined by special guest, Pete Hill @CudoPete, Co-founder and VP of Sales at Cudo (
) to talk about CUDOS, the significant growth in demand for decentralized GPU access, blockchain and AI.
Created to serve the AI market, CUDOS delivers scalable computing power to the blockchain ecosystem. With a global supplier network of GPUs built on 100% sustainable, renewable energy, CUDOS provides cost-effective and accessible compute power for AI training, inference, and machine learning workloads.
Volume’s first AI bot which uses GPU compute is currently in testing and integration. Llamalend bots for Curve and Uniswap v3 limit order bots are poised to go live soon.
Taariq welcomes everyone to Volume’s Twitter Spaces. He also extends a warm welcome to today’s special guest @CudoPete of @Cudos_ who will be introducing CUDOS later in the episode.
He shares a brief background about Volume.
The Paloma blockchain, developed by Volume, runs on the latest version of the Cosmos SDK and communicates cross-chain to eight target chains.
Llamalend bots, the trading bots on Curve continue to undergo testing. More automated bots for Curve’s Llamalend markets are in the brew.
Taariq congratulates all Paloma validators for completing $250k in transaction volume on Gnosis Chain and Ethereum for this week alone. All Palomabots are zero-fee for May and users only have to pay gas.
As Volume enters the AI space, Volume’s first AI bot, code-named Taylor, which uses GPU compute, is currently in testing and integration. The team continues to explore the blockchain and AI space and reach out to GPU providers for potential partnerships.
Paloma is a Cosmos SDK blockchain running at about 750 milliseconds block time with high throughput on Gravity Bridge. After the Bridge upgrade, Paloma will be able to execute bridge token transfers across the eight supported chains.
As the chain continues to upgrade Gravity and address CodeQL fixes, there are no new releases this week for Paloma.
Pigeon Feed, Paloma’s -marketplace for message relaying is how Paloma validators make money.
The host explains that gas payment on the target chain is monetized in GRAINS which creates buy pressure on the GRAIN token.
After the major upgrade to Cosmos SDK 50 last April, the team will be working on Gravity testing, including integrating multiple signing keys for Pigeons. This functionality will be refactored and tested into the new Cosmos before the GRAIN token is deployed to the target chains.
Taariq asks Pete to share a background about CUDOS, how it started, what inspired it, and their vision in decentralized compute and GPU access.
Pete shares that CUDOS is a UK-based Layer 1 blockchain built from Cosmos SDK focused fully on providing scalable computing power. Coming from a successful business as a data center cloud and connectivity company that grew into one of the largest data center aggregators in northern Europe, the initial vision was to democratize the cloud in the future.
As artificial intelligence began taking off, they realized that it would need an incredible amount of computing power while the consumption of cloud services continued to grow year after year. How can this power be harnessed in a more sustainable way?
He adds that with the AI boom happening, the amount of power and infrastructure required is exascale. A new company Cudo was created to serve the AI market.
CUDOS Intercloud Platform launched in January this year, which is the gateway into the interface of all scalable computer power. Roughly 700 GPUs are in the marketplace at the moment in different locations around the world. The backend supply chain is democratized with various suppliers worldwide providing hardware. The long-term vision is to provide a customer-edge device.
Taariq asks about the pricing at CUDOS and other decentralized DePIN GPU access providers compared to centralized players like Amazon, Azure, and Google.
Pete replies that coming from the cloud industry before, they witnessed the whole industry get torn apart with the market share going towards big hyperscale companies. These companies serve the enterprises, the banking industry, the insurance industry, and the government as well. Data centers are built next to major cities where the land price, electricity, and labor are higher, thus, costs are generally higher.
The actual AI training itself can be run in more remote locations which typically have access to renewable energy such as hydroelectricity, solar, and wind, which are generally lower-priced.
He explains that CUDOS Intercloud, through the underlying network, CUDO Compute provides a route to market with better-priced electricity and hardware, enabling them to pass on those economies of scale.
There are 700 GPUs in the marketplace available on-demand, pay-as-you-go arrangement built per second, but presented per hour. For those who want a longer-term commitment, or working from the traditional space in their training or language model, the network has H100s and H200s. The MI250s are coming soon, while the MI300s will be out later this year.
DePIN networks mining for machine learning or inference workloads use the CUDOS Intercloud to rent out the GPUs. They favor the A40s and the A600s rather than the H100s because of the price point making it more accessible to people. The A40 and A6000 are around 79 cents per hour and are purely on demand.
The CUDOS Intercloud supports multiple wallets such as MetaMask and Keplr, and multiple tokens such as USDT and USDC. For crypto payment, users can go to
to connect their preferred wallet.
CUDOS is actively working on incentives and utility for their token to drive adoption. Pete explains that the network had to pivot the supply strategy to the service provider environments because of the way AI needs to consume computing power. He adds that service providers and data centers do not typically accept payment in tokens.
There is a significant growth of new demand for decentralized GPU access after the launch of CUDOS Intercloud in January. From 200 users, they are approaching 1500 users four months on. While there is demand from the blockchain for validator nodes, for example, GPU consumption comes from those mining for workflows in AI, machine learning, and inference.
Taariq talks about an opportunity for Volume and Paloma to make it easy for people to use CUDOS tokens in an automated fashion. He cites how Paloma’s validator set can help reduce barriers to allow folks to use any token to pay for their compute power.
A part of the CUDOS vision is to scale use for multiple chains. CUDOS runs on the Cosmos chain like Paloma and has its own Layer 1. The strategy is to make computing power accessible not only to Cosmos but to the whole blockchain ecosystem, with the main goal for the CUDOS Intercloud to be available massively multichain.
Taariq expresses his appreciation for CUDOS.
He adds the prospect for Paloma to explore a partnership with CUDOS and participate in their network.
Pete states that CUDOS is always open to partnerships and finding ways to collaborate with more people and more projects in the space.
Taariq thanks Pete, the CUDOS community, Paloma, and Volume for tuning in to today’s AMA episode.
Stay tuned for the next AMA!
To find out more about Volume, check out Volume Finance (
), join the Discord (
), and follow us on Twitter (@Volumefi). Check out Paloma Chain on
(
), follow them on Twitter (@palomachain), and join the flock on Discord (
). Coo!