Filecoin Onchain Cloud Explained: Verifiable Decentralized Cloud for Web3 Builders

GM, GM. Welcome to what's going to be another fantastic episode of DevNTell. So if you didn't know, DevNTell is a 30-minute podcast held every week, allowing founders, hackers, and anyone in between the opportunity to come on the show and showcase what they built. And today, I'm ecstatic to welcome Sarah Thiam, who is Head of DevRel at Filecoin, who will be introducing us to Filecoin Onchain Cloud. So if you aren't aware, Filecoin Onchain Cloud provides transparent storage, retrieval, and payments on the Filecoin network. So if you stick around for today's episode, you'll get to meet Sarah, learn all about Filecoin Onchain Cloud, and how you can get started using it today. All right, let's get into it.

[Intro music]

GM, GM, or GN in your case, Sarah. Welcome to the show. Excited to have you on today.

Yeah, GM, GM. Thanks for having me, Narb. The intro video always gets me. It's such a mood setter.

Yeah, exactly. You've got to start with a bang, right? Be loom. But yes, it's wonderful to have you on the show, and thank you for taking the time out of your busy day to do so. I'm really interested, for myself and the audience here today, to learn all about Filecoin Onchain Cloud. I know it's a relatively new platform, so it's really exciting to kind of deep dive there. But before we get into that, would you just like to give an introduction about yourself?

Yeah, of course. Cool. Thanks for having me today. My name is Sarah, I'm the Head of DevRel for Filecoin. Filecoin is a network... it's now a blockchain, but it's also supported by a network of different companies and entities because we are truly decentralized in that way. So in a sense, I'm also a founder. I lead my own entity called Fil Builders or FilB, and we are the DevRel team for Filecoin. So we're a team of six of us and we've been supporting Filecoin for the past two years on our own, and prior to that, I was part of the Protocol Labs team before we became separate entities as a network, also doing DevRel at Filecoin. So that's my background. And before that, I was previously at Microsoft and also worked at the Singapore government. And yeah, it's been a long time in tech moving across the space from Web2 all the way to Web3. And now everything is becoming AI, so we're all there. I know how you feel. So that's kind of my background. Yeah.

Awesome, awesome. Yeah, it seems you've been well-traveled in the tech scene certainly and had your fair share of experiences. I guess, what was your kind of aha moment that you kind of said, 'Oh, yeah, tech, tech is the right step for me'? I guess what inspired you to get into tech?

I think it's... I think my first day into tech is probably not the best indication. My first day in tech was the day that Windows 10 launched and I was the program manager for it. So, you know, not indicative of why I chose to join tech. But I think it's just... I think I came in at a time when cloud computing was becoming really big. So for those that are going through the AI boom right now, if you were around during cloud computing, you know what I mean. It was probably a smaller scale, but it was also the word on everyone's mouth, like 'oh, cloud computing'. But back then I was like, hey, tech is going to be everywhere, it's not going to be an industry on its own, the same way that AI is going to be everywhere and not an industry on its own. And I just thought it made a lot of sense to jump into it and get to understand, in a sense, the fundamentals of something that everyone's going to be using. Because everyone always turns to you and is like, 'how does this actually work?' And you see the applications of it. And to understand what's going on under the hood and the infrastructure behind it and what actually... what are the privacy, security, cost, and all of these other considerations is super interesting. So that's how I chose tech as my career and I've never really looked back.

This is the way, to steal a line from the Star Wars universe. I feel like you've probably been in tech the same amount of time as me. I think it's probably about 10 years, right?

Yeah, exactly. Yeah, about 10 years, maybe 11. But yeah, it's... yeah, I've never, I've never looked back. I don't know how I would survive in any other spot but this. So, yeah. And I guess you mentioned it, from Microsoft, you made your way to the Filecoin Protocol Labs universe. I guess for people who are watching today who might be new to the blockchain side of the tech spectrum, I guess what would be your high-level kind of overview or... definition of kind of what decentralized storage is, kind of setting the scene for Filecoin itself?

Yeah, that's a great question. Decentralized storage is actually a really interesting concept. It's also one of the reasons why when I pivoted from Web2 into Web3, I actually... so I was very fortunate, Protocol Labs was my first choice and I got in. I was very, very lucky. I did not expect to be able to work with such amazing people and such excellent engineers at the time. And the reason why I chose it was because I looked across the Web3 space and what attracted me to Web3 and blockchain was the idea of sovereignty. And working in FAANG before, also being very fortunate there, you kind of see how it's a huge machine and it's great, but it's also a few companies making a lot of decisions that influence a large part of the tech sector. And, you know, a lot of us know this firsthand, right? Like there's not really many options out there. And so the question then came up with, like with cloud computing and a lot of data going into the cloud, like who actually owns this data? And I think in the past, even in the past six months, I think in Q4 last year, when AWS went down, a bunch of servers went down, a bunch of websites went down, everybody was affected. And these things happen from time to time. And if it happens on a larger scale or if it happens for a longer time, then that can really impact a lot of people's infrastructure, lives, everything, right? So it is very important to have decentralized infrastructure sitting somewhere, even though for various reasons, usability and these things, like these things are still progressing. We all know this in the blockchain space, it will take time, but it's good to have alternatives. And so with data being so important, storage is definitely a key component to the stack. And I looked at the options out there and I was like, you know, decentralized storage in itself is just important for censorship resistance, that's really important. When governments go down, you can see like during times of crisis, there are certain nations or countries or situations where people have used different applications like BlueSky or... which BlueSky actually uses IPFS and Filecoin, which is awesome. And that's what enables them to communicate with each other and spread the message, right? And so you can see the importance of this. And so that's what really attracted me to decentralized storage and the importance behind it. Yeah. I mean, it's a lot, it's a lot there. I can keep going on, but that's why I was like, this seems like cool stuff that needs to get done.

Yeah, certainly. And when I was first getting into the blockchain scene, I mean, I started with the Bitcoin white paper and Ethereum white paper as well, but then what really caught my eye, it was decentralized storage. Like, I got into Arweave, I got into Filecoin. I just thought the concept was like really cool. IPFS as well, I know it's not exactly storage, but it all kind of works together. And you mentioned it briefly there, we have seen a lot of the AWS's, the Azures, what have you, kind of going down a little bit more frequently than what I recall in recent memory. I think whether or not it's attributed to AI or maybe people are using these agents in nefarious ways to hack or DDoS these platforms, but yeah, definitely having these decentralized alternatives seems very important, especially in this day and age.

I think 100%. I think maybe what you mentioned is probably a good hint at maybe why these things are happening more often. I think, and I don't think it's anyone's like or any infrastructure's like, you know, flaws or anything, it's just it's an unprecedented amount of data that is being generated with AI, both on like the amount that is needed to run the AI and the amount that the AI outputs. And if you think of the current options that we have, it's just not going to be enough at the exponential speed that it's growing. And we will need like, you know, if everyone... if multiple like people's or, you know, let's just say a common man's computers need to be used for data as well or storage, then that might even become an extremely important option to have. So, yeah, even more so today.

Yeah, exactly, exactly. So I think this was a good segue. So when I first started learning about going into the Protocol Labs universe, obviously there's IPFS, there's Filecoin, and then the Filecoin Virtual Machine. But it kind of seems like now this is all kind of come together in a single place and making it easier for developers to kind of get access to this. So yeah, I'll let you take the reins and tell us all about Filecoin Onchain Cloud.

Yeah, um, where do I begin? I feel like I've been in Web3 for like five years, but it's only been like three. And in that time, I've seen... I think I've been in a really privileged position to see how Filecoin has evolved, actually really a lot. Like the day that I joined Protocol Labs and the Filecoin project was the day... was the few months right before the Filecoin Virtual Machine was going to be launched. Back then this was like late 2022, early 2023. And what Filecoin is really aiming for, first of all, let's just start with IPFS-Filecoin, and I'll try to be as high level as I possibly can, but I do think that the lore is like interesting and important context for how we arrived at Filecoin Onchain Cloud. And also even until today, I think people do think... and it is a very understandable confusion, maybe a mistake on our part, but people do think that IPFS is Filecoin and they are interchangeable. And like you said, Narb, I think you're exactly right, they are not the same. IPFS is a peer-to-peer network, it's not even... it's not blockchain. It does allow you to store, but more like cache, and then retrieve it really quickly in a peer-to-peer network. But it doesn't mean that your data is guaranteed to be stored forever unless you're using like a guaranteed paid service like, for example, Pinata and so on, right? Or Infura. Filecoin on the other hand is backed by a blockchain and it is the... it is inspired by IPFS, it uses the core technology that IPFS uses, like Merkle Trees and PCIDs to be able to store your data and make sure that there's data integrity, but there is also the backing of an entire economic network of storage providers secured by the blockchain so they cannot actually lose your data or they will get slashed really, really badly and penalized. So that is the core difference. And even until today, people are not very clear on the difference. They do work hand-in-hand. Filecoin does use IPFS, IPFS does use Filecoin. Filecoin Onchain Cloud does change a little bit of that as well and helps to merge the two together. I do remember last year in Argentina, we got Filecoin Onchain Cloud pre-launched during Argentina DevCon. And at one of our Fil Dev summits, Juan was like... Juan Benet, who is the founder for IPFS and Filecoin, was in the room and he was like, 'finally after six years, IPFS and Filecoin can finally work together'. Like things from IPFS can be backed down to Filecoin. And we were like, 'oh my gosh, yeah, it took us six years, what?' So that's... and I'll allude to that, I'm alluding to the Filecoin Onchain Cloud, but that's how we got there. So Filecoin-IPFS separate, Filecoin came about because IPFS saw the need for economic guarantees. Then after about four years of Filecoin, three years of Filecoin running, we saw a need to be able to run smart contracts on Filecoin because people were asking, like, 'we can't go around making storage deals by manually contacting through phone or email a storage provider and being like, hey, can you store my data? Do you? Like yeah sure'. Like that's just not going to cut it, right? So it has to be programmable to be able to be scalable. And so retroactively, we built a virtual machine that was... that is EVM compatible, went into Filecoin. And so that's how we came with the FVM. And so that has... that is also the basis for why we have the Filecoin Onchain Cloud today, or are able to have it, because you could finally run smart contracts and deploy them on Filecoin. And FVM was really cool, but also we were held back a little... it ran for about two years, we had a ton of builders, we had a lot of different people building applications. I myself was heading the team that ran three to four early builder cohorts, and so we had a bunch of builders that built with us for six to nine months and went really, really far and have continued to build, which is awesome to see. But also we received a lot of good feedback that, like, it's just still not fast enough. Because Filecoin as a network was originally built with... I think we also were the pioneers, one of the pioneers to have ZK proofs way before ZK proofs were a thing. And we use ZK proofs within the storage proofs that we have on Filecoin blockchain. So storage providers will be required to provide a ZK proof, which is computationally extremely heavy and it takes like hours for them to complete that challenge to show that they have the proof to show that they have the data available. And that led to storage deals being like, it would take 24 to 72 hours, even with the FVM. Like you could programmatically set it and be like, 'store my data at 12 o'clock' and it would store your data, but then you probably have to wait three days for your data to be stored, which is like, you know, back in pigeon days when we're sending pigeons with letters, which is what... no, we shouldn't be that, right? So you're like, 'okay, we definitely need to fix this'. Retrieval is equally painful. So right now the only use case that we have is for really, really large data, and mostly cold storage. And we have worked with many institutions, we've worked with the MTA in New York, we have several use cases, we have Berkeley, we have a few other universities and storing a bunch of institutional data, but it's meant for archiving and not really meant for them to see it at any one point in time. And so we wanted to shift into like, how can we allow developers to use storage and retrieval, which is what they really want, on the go and really, really fast? And so we started designing, I think sometime early 2025, late 2024, we started designing the Filecoin Onchain Cloud. This was led by our cryptoeconomic lab within Protocol Labs, one of the researchers, Nicola, he came up with the thesis of like, we should have Filecoin Onchain Cloud, think of it as a marketplace with a bunch of storage retrieval services within that marketplace that run as smart contracts and it's all modular components. And in the future, you know, Filecoin can do what Filecoin does best and provide storage and retrieval first, and if there are other options in the future for like compute, smart contracts, imagine, so you are able to run compute next to the data that is being stored, like a single storage provider could do data and compute physically where the data is being stored, which would reduce cost and speed drastically. So that was the vision. And so the FilOS team, which is the core dev team within Filecoin, again, we are many different entities today, so that's the FilOS team, and that's run by Molly and Jennifer, and they have been building, doing some incredible work with the engineering team, they've been building out the Filecoin Onchain Cloud. And that is where today we have storage and retrieval ready, and that has allowed us to not only two things, two very key things: the first one is to store data on a much smaller scale, so there is no longer a minimum limit of like 4GB of data, which developers for the two years that they were building struggled, really struggled to be able to hit 4GB to store, which is insane. That's like a movie, that's like a 1080p movie, and they just wouldn't store that, right? So that has been reduced, you can now store a minimum of 127 bytes, which is a lot smaller. And at the same time, storage... we actually came up with a new proof format that is very similar to the previous ZK proof that I mentioned, because that actually interacts with the original consensus, whereas now we have proof of data possession, and I know I'm saying a lot of words, but TL;DR is like, we have new proof format that's a lot lighter and the same storage providers that have their physical hardware can now provide much lighter proofs, it's a lot faster, people with even smaller amounts of data storage can actually participate as storage providers so we can have more, and that allows your data to be stored much quicker. Like, we are down to like seconds now rather than the hours I mentioned before, which is insane. 72 hours down to like... I think the fastest we got close to just under a minute, and we're still improving that drastically, which is like, best improvement award. And retrieval, we have a CDN now, a custom CDN that has been built called Filebeam, and that allows you to retrieve your data in under 10 seconds, and we're continuously reducing the time to first byte. And that is also complementary with the... as part of the Filecoin Onchain Cloud. So all in all, you're able to small... store smaller data and also store it and retrieve it really, really quickly. I can talk more about cost, but TL;DR, that's the evolution to the point of me seeing FOC being launched in 2026... 2025, 2026. We just went to mainnet last month, so that was really huge for us. But yes.

That's amazing. That's quite the timeline. And yeah, I remember just like watching as all the different solutions around the Filecoin ecosystem kind of started to evolve. Everything kind of seemed at that time like to be its own kind of thing, but now it was all puzzle pieces kind of coming together and I think this is awesome to see. And you alluded to it before, Sarah, I think it'd be awesome to kind of get a high-level overview of what the architecture of FOC looks like.

Yeah, definitely. So there are a few... as I mentioned earlier, the FOC stack is intended to be a modular marketplace of several smart contracts that allow you to put bits and pieces together to build your own custom solution, whether it's and it's all data related, if it's storage only, retrieval only, storage-retrieval, if you want to build your own compute, you can, and these should all work together. But there's some core components within the FOC stack. The first one being Filecoin Warm Storage Service, the name might change, but right now it's called FWSS. This is the main smart contract that allows you to store your data. So this fully just focuses on taking your data, prepping it, storing it, and sending the proof... sending it as piece of data to the storage provider and then receiving back the proof. So this is what that main storage... that FWSS smart contract is for. So that's the first one. Then that's store. So the second one would be around pay. So we have a smart contract called Filecoin Pay, and this is also a very key contract, arguably even more important as the backbone of FOC rather than the storage contract because this is all about your payment and your money and where it will sit. So what it allows you to do is to hold your own wallet. The way that FOC works, we've also designed it in a way that it only pays upon verification. So only at the point when the storage provider returns the proof of storage or the proof of data possession, it's called PDP in FOC, only then can they receive the payment and only then does the customer's funds actually get paid across. So until then, if nobody sends the proof back, you're not actually paying for anything, you don't pay anyone, you don't pay us, we don't handle anything, it goes through the Filecoin Pay storage contract. So that is the Filecoin Pay storage contract that's pay. And then the third component that I'll talk about is not... it's not a smart contract, this one is a CDN that we have. So I mentioned earlier to Narb that because Filecoin wanted to reach the speeds that were requested by developers, we actually had a team that was building Filebeam. The team is now part of FilOS, but this previously used to be a separate team and now they've become part of FilOS as well and it's called... the CDN itself is called Filebeam. And so the interesting thing is that it actually allows you to retrieve your data at much lower rates compared to other alternatives out there in the market. It's actually pretty comparable to AWS and slightly cheaper and it's constantly improving its speed. Like currently it's about one second to retrieve a small piece of 1MB, I think 10MB data, but we're getting faster and faster and even lower over time. So those are the three components: store, pay, and retrieve in FOC as of today, and we're continuously trying to expand that. Right now we're really interested in looking into compute options for how we can add that as an additional component to the marketplace. We're actually looking out for partners that we could work with and we would love to see even more use cases come out from builders on how they want to use storage and compute, but that's as of today, that's the current stack.

Awesome, awesome. And yeah, it certainly seems like there's a nice set of puzzle pieces, a formula if you will, kind of coming together for people to start building agents on with FOC. I guess, from your perspective, have you kind of seen a rise in interest in building agents around Filecoin?

Yeah, um, I think we are... honestly, we have been really happy with the reception when it came into the agent side of things. So, you know, AI boom and all, and then people started talking about, 'oh, what does the decentralized AI stack look like?' And I think... I think we're... well, shoutout to Nader if he ever sees this, but thank you for tweeting about Filecoin-IPFS as part of the dream, the dream team, DeAI or decentralized agent stack. So Filecoin-IPFS is very well recognized. I think if you ask builders in general in Web3, decentralized storage-wise, who would you pick, most of the time the first answer that comes is Filecoin-IPFS, and we're glad to have such brand recognition. Of course, we back that up with the stack that we have available today, but we do see many builders interested in building agents. We actually have a really cool demo using Filecoin Pin, which I don't think I have time to show, but I can very quickly talk about it and it actually is very quickly executed as well. So if you are using ERC-8004, we've been partnering very closely with the team, we actually... I think in Argentina we were part of the Agent Day, and then also recently we did the Synthesis hackathon with EF and also was very closely working with them for their launch. But if you're using ERC-8004, which is Agent Registry to register your agent on the blockchain to register its identity, you can actually take the metadata URI of your agent card and you can actually store that to IPFS before you... if you can store to IPFS using Filecoin Pin. Don't get confused if I want to say IPFS. The... I'll quickly talk about Filecoin Pin, which is opening up yet another branch of lore, but very quickly Filecoin Pin is like... I mentioned earlier that IPFS and Filecoin is always dreamed of being merged together. Of course, they are two separate technologies, but with Filecoin Pin, it was actually an application that we wanted to use as a demo for the FOC stack. And so the first thing we thought of was, how do you connect an IPFS interface or people using IPFS, have it pinned onto an IPFS node to be cached very quickly, but to have it also automatically backed up to Filecoin at a very, very low cost such that all your IPFS storage can actually be guaranteed and stored permanently on Filecoin as long as you pay for that storage. So that's Filecoin Pin. But as I go back to your agent stack, if you're using your agent card, you can store it onto IPFS and Filecoin using Filecoin Pin, which is very easy to use as a CLI. Again, you could use any of the resources I mentioned earlier as well and then you could instantly register that and your data, your agent card is basically registered and stored forever. So your agent never really disappears and you don't really have to pay that high an amount, it's almost negligible. So yeah, I think that's like the dream case and I think many developers are very excited about that. Other things that we're looking at AI as well, like we do see many devs when we see... we ran Synthesis recently, we're doing a few more hackathons coming up, and I think a lot of the use cases that we see is to store agent metadata, so agent logs are also something that's being stored, episodic memory for agents, maybe also like the context of the conversations, the outputs, still pretty light, still within the KB to GB range, but all this is being stored on FOC today. And I think further we're looking into the market for AI around like vector databases, which is larger and larger amounts of data that we expect. There's also a very healthy ratio between hot storage and cold storage for vector databases, so we think that we could really play a big part there because Filecoin has like, I think it's the largest decentralized storage network in terms of storage capacity today, we have like exabytes of data available. So we're really interested into this use cases and then also synthetic data is also something that we're looking into because it also touches on the hot storage and the cold storage and from small to large pieces. So I think these are some of the use cases that we're very interested in for AI.

Awesome, awesome. And I guess, are you guys looking for contributors to help build that out or any specific like grant programs for people to kind of get involved?

Uh, yeah, I'm actually working on a grants program right now. Um, so I guess it's on me. I'm also currently on vacation too, like yeah give me... give me like a week. Yeah, I'll come back. But we're working on something right now. I think what we think is very interesting about this grants program is that... and I think in the... this is less Filecoin-centric but more like general space for developers is grants usually went to building entire applications, so more build-focused. But now we're actually very curious about how can we be more GTM-focused with teams given what AI has transformed for everyone. Of course, we still want really impactful use cases, we still want a high level of technical skill and understanding, but we know people can build faster than ever and we're more interested into traction. So we're looking at milestone-based traction and how best we can support you through the grants program if you're building on FOC. So do look out for that, we will be keeping an eye out for any teams that participate in the hackathons that we're running. We just ran FIL Genesis and we just finished Synthesis. We're looking at a few more coming up, so follow stay tuned on the Filecoin account if any hackathons are coming up. And we... my team six people, if you win we'll see you. We'll know who you are. If you apply for the grants, we'll also know who you are. So, yeah, just stay locked in and we will definitely take note of you and if you're interested, come talk to us and we're more than happy to help you grow.

Wonderfully said and yeah, you answered my last question as well. So there you go. That's where you can follow Sarah and all the great stuff Filecoin and the FilOS network is kind of building. So definitely get involved. Filecoin, like we've mentioned and you heard in the lore, has grown quite a bit since early days, so no better time than now to get building. And with that, Sarah, thank you so much for taking the time out of your vacation to come talk to us today. I really enjoyed our chat and I know our audience did too.

Yeah, ditto. Thanks so much, Narb. I did very much too and yeah, thank you for having me and thank you for having us.

For sure, my pleasure. And yeah, with that, I just want to wish everybody a very happy Friday, happy weekend wherever you may be, and we'll catch you back here for another great episode next week. Until then, have a good one, folks. Cheers. Bye.

Bye.