Pinned Tweet
We are Shelby!
First ever decentralized cloud-grade data layer!
First ever decentralized cloud-grade data layer!
It's time to challenge the cloud monopoly.
Meet Shelby, the first decentralized cloud-grade infrastructure for monetizing and serving real-time data.
Incentivized reads. Sub-second speed. Any chain, any stack. Web3's data solution has arrived.
shelby.xyz
10
9
100
13,343
Yes, blockchain coordinated infra can match cloud performance.. Shelby private beta is humming along and have already done a sustained load of
> 1000 requests/sec
> 1-2 TB/Hour/customer
Barely making a dent in total system capacity.
> 1000 requests/sec
> 1-2 TB/Hour/customer
Barely making a dent in total system capacity.
12
17
63
5,918
pranav | Shelby retweeted
Shelby is built for workloads where:
→ compute moves between locations, providers or environments;
→ data is collected, processed or trained in different places;
→ new compute creates more staging, copying or pipeline work;
→ the same dataset needs to serve multiple workloads.
Sound familiar? Get in touch.
→ compute moves between locations, providers or environments;
→ data is collected, processed or trained in different places;
→ new compute creates more staging, copying or pipeline work;
→ the same dataset needs to serve multiple workloads.
Sound familiar? Get in touch.
28
32
145
19,236
Quick onboarding, strong partner validation.
Shelby is just starting.
Shelby is just starting.
Can vouch for the product and team.
Had a what I thought was a complex storage need for @BaranosAI and was onboarded in less than 24 hours. ✅
Don’t sleep on @shelbyserves
3
2
31
937
Why use Shelby?
"We can switch GPU providers without revoking the data layer every time"..
@teepin has lined up a few enterprise customers. Very excited to work together!
"We can switch GPU providers without revoking the data layer every time"..
@teepin has lined up a few enterprise customers. Very excited to work together!
Teepin is building on @shelbyserves Private Beta.
We are one of the first projects running real workloads on it.
Learn what we are building on Shelby and how Shelby enabled our business model 👇
6
5
42
1,358
pranav | Shelby retweeted
3 enterprise-grade AI customers live on @shelbyserves.. AI data will power trillions in GPU spend. A global namespace closes the gap.
Shelby Private Beta is now running real customer workloads.
We're starting with a small group of teams across enterprise AI, spatial intelligence and distributed rendering.
What we learn together will shape what comes next ↓
11
11
90
2,993
Shelby is live!
Eight DCs across US and EU. > 500 TB of data stored in a month. 100s of concurrent uploads/downloads. Lower subsecond TTFB.
We have been head down building a global data network coordinated on a high performance blockchain, free of cloud silos, regional buckets, egress taxes. Plus cryptographic verifiability baked in. Single address with global access.
Customers are live and validating these claims. Developer access coming real soon.
Eight DCs across US and EU. > 500 TB of data stored in a month. 100s of concurrent uploads/downloads. Lower subsecond TTFB.
We have been head down building a global data network coordinated on a high performance blockchain, free of cloud silos, regional buckets, egress taxes. Plus cryptographic verifiability baked in. Single address with global access.
Customers are live and validating these claims. Developer access coming real soon.
Shelby Private Beta is now running real customer workloads.
We're starting with a small group of teams across enterprise AI, spatial intelligence and distributed rendering.
What we learn together will shape what comes next ↓
24
28
94
10,796
pranav | Shelby retweeted
What we learned from Shelby Testnet ↓
Testnet gave us confidence in the core platform.
Putting Shelby in the hands of builders gave us direct feedback on the tooling, APIs and compatibility they actually needed.
Testnet gave us confidence in the core platform.
Putting Shelby in the hands of builders gave us direct feedback on the tooling, APIs and compatibility they actually needed.
57
46
251
30,953
pranav | Shelby retweeted
Moving data has a cost. So does waiting for it.
Transfer and egress land on one bill. Compute waiting for data lands on another.
Price your data path against your compute bill, not just your storage bill.
Transfer and egress land on one bill. Compute waiting for data lands on another.
Price your data path against your compute bill, not just your storage bill.
27
41
149
7,534
Real customers, high throughput environment, multi-continent footprint.
AI compute is already global. Shelby is now breaking data silos..
AI compute is already global. Shelby is now breaking data silos..
We’ve been heads down hardening Shelby for Private Beta, and we’re beginning to onboard a select group of customers in stages.
Over the next few weeks, we’ll share more about what testnet clarified, how those learnings shaped the product, and the problem Shelby is built to solve as compute becomes more distributed.
14
4
47
2,120
Shelby coded.
Not your keys, not your data.
Not your keys, not your data.
The goal of Personal AI: civilization where individual humans, augmented by AI, can do consequential work without being captured by extractive institutions.
Freedom to write your prompt and own your data.
This is the new battleground.
2034 won’t have to be like 1984.
12
4
33
2,046
pranav | Shelby retweeted
@MilkRoadAIcarcity, acquiring compute wherever you find it has its own challenges. Once you acquire it, you need to ensure you can orchestrate your workflows and data across the compute clusters. Most AI labs and Neoclouds are setup single DC only. This is highly inefficient. @shelbyserves aims to bridge this gap.
1
2
9
654
pranav | Shelby retweeted
AI agents that transact need more than fast compute. They need data they can prove.
@rpranav is breaking down what that means for payments, tokenization, and the next layer of commerce at @googlecloud Next tomorrow.
With @chandler_agi, @sreeramkannan, and @jmtromans.
@rpranav is breaking down what that means for payments, tokenization, and the next layer of commerce at @googlecloud Next tomorrow.
With @chandler_agi, @sreeramkannan, and @jmtromans.
40
40
221
15,992
Shelby coded : - use blockchain to solve a real problem.
Shelby Global Namespace - only possible due to blockchain as a coordination layer.
Data rails for agents - verification of ownership, payment, data quality.
Much cheaper than S3 due to decentralized OpEx.
Shelby Global Namespace - only possible due to blockchain as a coordination layer.
Data rails for agents - verification of ownership, payment, data quality.
Much cheaper than S3 due to decentralized OpEx.
@theSamPadilla Totally fair question. Anything that solves a user problem and just happens to use blockchain bc its cheaper/faster/more efficient etc is the future. Speculative businesses or ponzis are out.
13
4
38
2,262
pranav | Shelby retweeted
If you're offering compute but egress is eating up your margins, let's talk
If your training data is stuck in one region and your GPUs are in another, let's talk.
If your content is being consumed, but you don't know by whom, let's talk.
Booth E11. We're here all day.
If your training data is stuck in one region and your GPUs are in another, let's talk.
If your content is being consumed, but you don't know by whom, let's talk.
Booth E11. We're here all day.
32
16
156
5,578
A fun way to explain a huge roadblock in AI growth story.
I was talking to one of the biggest cloud providers last week. This is THE top problem for their customers. No GPU available in the region customer has data in. And the cloud provider doesn’t have a global namespace.
Shelby is building a new data paradigm.
I was talking to one of the biggest cloud providers last week. This is THE top problem for their customers. No GPU available in the region customer has data in. And the cloud provider doesn’t have a global namespace.
Shelby is building a new data paradigm.
The greatest love story in AI infrastructure.
1
6
48
4,198
pranav | Shelby retweeted
"Just put a CDN in front of it."
CDNs cache popular content. Great for viral videos.
Terrible for AI training data that gets accessed once per experiment.
Terrible for long-tail content nobody's requesting twice. Terrible for write-heavy workloads.
A
CDN is a band-aid on top of regional storage.
@shelbyserves eliminates the regional storage problem entirely.
Your data has no origin region. It just exists everywhere.
CDNs cache popular content. Great for viral videos.
Terrible for AI training data that gets accessed once per experiment.
Terrible for long-tail content nobody's requesting twice. Terrible for write-heavy workloads.
A
CDN is a band-aid on top of regional storage.
@shelbyserves eliminates the regional storage problem entirely.
Your data has no origin region. It just exists everywhere.
4
1
24
1,279
Always a blast talking to @Austin_Federa ..
Shelby makes data mobile, verifiable and cost efficient. While being
A single Global namespace to move data anywhere.
Cryptographic proofs for ownership, existence, payments, access.
Fraction of the big cloud cost.
Same performance and reliability.
Shelby makes data mobile, verifiable and cost efficient. While being
A single Global namespace to move data anywhere.
Cryptographic proofs for ownership, existence, payments, access.
Fraction of the big cloud cost.
Same performance and reliability.
Shelby, critical infrastructure for data-intensive applications, is now live for early access.
In this episode, @Austin_Federa sits down with @rpranav to break down @shelbyserves architecture, market opportunities, and more.
Full episode ↓
8
3
48
3,685
pranav | Shelby retweeted
Power, compute, & data co-location are essential for AI. And moving models and training data around is now the biggest tax on innovation.
When we started the collab with @jump_ last year, I believed @shelbyserves would be the most powerful & disruptive serving layer imaginable. I didn't know how important it would be to teleport data around for training, inference, fine-tuning, long-term memory, and more that spans a global infrastructure. Its scalable nature localizes data to every company and person in the world at ~70% less cost than the big hyperscalars, with higher availability due to decentralization.
@shelbyserves early access today is a huge milestone on this journey. There has been tremendous outreach from so many builders who have experienced Shelby devnet and know that there is finally a product that fits their needs. A high-performance storage serving layer takes time to build (I've done this a few times in my career, lol). Shipping a novel product this quickly is a direct result of Jump/Aptos talent & synergy.
This type of high-performance serving infrastructure needs the high-performance trading engine in @Aptos (<40ms block times, 99.99% uptime, native trading VM). We're blown away by the builder community creating on devnet - large enterprises and startups alike. Testnet is a bigger, better playground, mainnet coming soon!
When we started the collab with @jump_ last year, I believed @shelbyserves would be the most powerful & disruptive serving layer imaginable. I didn't know how important it would be to teleport data around for training, inference, fine-tuning, long-term memory, and more that spans a global infrastructure. Its scalable nature localizes data to every company and person in the world at ~70% less cost than the big hyperscalars, with higher availability due to decentralization.
@shelbyserves early access today is a huge milestone on this journey. There has been tremendous outreach from so many builders who have experienced Shelby devnet and know that there is finally a product that fits their needs. A high-performance storage serving layer takes time to build (I've done this a few times in my career, lol). Shipping a novel product this quickly is a direct result of Jump/Aptos talent & synergy.
This type of high-performance serving infrastructure needs the high-performance trading engine in @Aptos (<40ms block times, 99.99% uptime, native trading VM). We're blown away by the builder community creating on devnet - large enterprises and startups alike. Testnet is a bigger, better playground, mainnet coming soon!
1/ AI needs crypto's infrastructure. Crypto needs AI's scale.
Shelby is built for both. Early Access is open. Apply to build today.
17
22
103
5,962
pranav | Shelby retweeted
With changing compute dynamics, modern AI workloads need data served at the edge. @shelbyserves is the high-performances storage layer for modern edge AI — a highly cost-effective global namespace and verifiable object store with sub-second reads, parallel data flows, and programmable access so models can stream and serve data at cloud speed, anywhere. Early access now open for developers to test !
shelby.xyz/article/shelby…
shelby.xyz/article/shelby…
3
4
34
1,771
Write once, read anywhere.
It is a very powerful primitive. Bringing your data to compute, on demand, without replication is a game changer.
Done right, decentralized infrastructure has tangible benefits.
It is a very powerful primitive. Bringing your data to compute, on demand, without replication is a game changer.
Done right, decentralized infrastructure has tangible benefits.
The more I learn about AI infra, the more shocked I am by how much money we're burning on moving data around rather than on actual compute.
Your training data sits in one region. Your GPUs are in another.
Every time you feed your model, you pay egress fees. (cost for data leaving the database)
Want redundancy? Duplicate your dataset across 5 regions, paying 5x the storage cost.
GPUs are scarce. You can't always choose where your compute runs. So your data has to travel to the GPUs, not the other way around.
That means $$$ in egress costs.
That's why I'm genuinely impressed by what @shelbyserves is building.
They're not patching one piece of the problem. They're going after all of it:
➡️ Egress fees? ~70% lower.
➡️ Regional silos? Single global namespace: write once, read from anywhere.
➡️ Data provenance? Cryptographic proof on every read.
➡️ Vendor lock-in? Decentralized. No single provider owns your data.
The unnecessary cost that AI companies absorb today because of inefficient infrastructure design is staggering.
Shelby is building what this infrastructure should have looked like from the start.
6
1
32
977