Full Transcript — From Cloud Monopoly to Open Infra: Who Wins the AI Stack?
Industry leaders from Render Network, Manifest, RenderLabs, Jember, THINK, Virtuals, and Scrypted came together for a wide-ranging Spaces discussion on the future of AI infrastructure. Hosted by Sunny Osahn (Render Network Foundation) and Player1Taco (Manifest Network), the panel explored how decentralized networks can challenge cloud monopolies, unlock idle GPUs worldwide, and create open systems where value flows to builders and creators across the AI stack.
You’ll find the full recording and transcript below.
Sunny Osahn (the Render Network Foundation):
Good morning everyone, I hope you’re all well and good. I think we’re going to give this space maybe another minute or two just to make sure all of our guests are here and that they have the ability to speak. Yeah, so let’s do that. Thank you for the thumbs up there, Rena. And I assume everyone can hear me because I just got a thumbs up. That was that’s always my main question, first of all. So yeah, let’s give this
03:25
a minute and then I’ll get clarification that we have everyone here. Edward has just joined. We have Don from Virtuals. Amazing. Right. Back in a second, guys.
04:31
It is eerily silent here, isn’t it? So how about we just talk about what we’ll be talking about. I do believe we have the majority of our guests here. Let’s just double check. I know our guests are very eager to be speaking on this. We have a very interesting one for you guys today. You know what? Let me just give you guys a bit of an intro. So my name is Sunny and I’m
04:59
from the Render Network Foundation. The Render Network Foundation facilitates the strategic vision for the Render Network with the goal of growing its adoption through various strategic initiatives. It does so by engaging with the community and stakeholders from creators and digital artists to GPU node operators and ecosystem partners building on the network. And I am absolutely delighted to be hosting this space today alongside my friend, Taco.
05:29
Player1Taco, Manifest Network
What’s for you? Yeah, hi Sunny. I am doing well. That, I’m going to call it out. That was the most corporate complex way to say, we give creators really cool thing, tools to create cool shit. You know what? I would love to just jump on the space and say, listen, guys, we build shit and we’re good at it. You know, we love artists and I was going to go into a bit more depth with some of the artists, know, any award winning and stuff like that. But you know what?
05:59
Sunny Osahn, the Render Network Foundation
I’ve been told not to blow my own trumpet. My wife tells me not to do that quite often. but I will blow it for you. Render that. I’m to a hot start today. No, one of the really great things I love about render is this artist showcase and the competitions alone and how art that brings out.
06:27
on the 3D worlds, on the 3D art, and especially the recent showcase of render around the world has been really awesome. Yeah. Yeah. You know what? We have quite a bit planned and we will be talking more about this in the coming weeks. Yeah. I’m extremely pumped for…
06:50
the next few months. It’s amazing. And as always, we have some of our regulars listening as well, as well as new people who may not know about us. So welcome everybody. We are very, very glad you’re here to talk about AI with us, or at least listen to us talk about AI. The title of this space is in particular, From Cloud Monopoly to Open Infra, Who Wins in the AI Stack?
07:21
So essentially in today’s AI economy, value increasingly accrues to centralized providers, those who tend to control the compute, the models, and the delivery layer. But an alternate paradigm is emerging, one where open decentralized infrastructure spreads value across the stack, from GPU providers to AI app developers, even to creators.
07:46
Whoever cracks distribution of services and workloads will ultimately shape the future of AI. So this is what we’re going to be talking about today with some of our amazing guests. We have various people. I’m sure some of you have already read the tweets that we’ve put out. We have some of our ecosystem partners, our friends from RenderLabs. have Danny and Paul from RenderLabs, which is absolutely amazing. You know what?
08:13
How about I actually let you guys intro yourselves? So one by one, you know what? I’ll call upon you guys. Basically, who are you and why are you here? What is it, which organization are you part of, and what do you do? So let’s start with Danny.
Danny Newman, RenderLabs
Hey, everybody. I am Danny Newman, part of the RenderLabs team. RenderLabs, we ideate, incubate, and launch innovative products.
08:43
and businesses specifically optimized for the render network. So we are working on a couple of projects internally. We are working with groups, teams, companies externally to bring more compute projects to the render network. Excellent. Thank you so much for that intro, Danny. Paul? Yeah.
09:08
Paul Roales, RenderLabs
Good morning. Good to be back here with everyone. It’s always a fun hour. So I’ve been looking forward to this all week, but yeah, similar to Danny working on RenderLabs and, you know, been working on ML stuff for 10 years across Waymo and Google and excited to these days be working on our decentralized network at Render to bring excellent experiences to that. And, you know, just explore and expand what can be done. Very exciting days. So yeah, thank you.
09:37
Amazing. Thank you for that, Paul. We also have Edward from Jember.
Edward Katzin, Jember
Thank you, Sunny. It’s a delight to be with you all again. Yeah, Jemba, we’re pleased to be working with both Render and Manifest Network, deploying our infrastructure, deploying services, and helping to bring these visions to life. Excellent. We also have Mike from THINK.
10:03
Mike Anderson, THINK
Hey, thanks so much for having me. So what we’re doing over at ThinkAgents is we’re standardizing consumer demand. What does that really mean? It means we’re giving the agents an on-chain token that acts like their address. It’s almost like if you bought a .com during the, you know, to have a website hosted, this token gives your agent a name. It gives it the authentication so that it can, it doesn’t have to deal with APIs. It can do it all via Web3 signatures. And this agent now we’re getting ready to release the ThinkOS browser.
10:32
And essentially what this does, it allows anybody to have a browser based on the Chromium standard that actually has THINK Agents built into it. So whether it’s your agent or an agent that you buy or rent through the system, you will have a common way to be able to interact with these agents. And that’s the foundation of an agent economy that’s coming. Thanks for having me. Amazing. Thank you so much for that. We also have Don from Virtuals. This is the first time you’ve been on any of these spaces that we do.
11:02
So I’d love for you to intro yourself.
Don Johnson, Virtuals
Yeah, guys, thanks for having me. I’m Don Johnson from Virtuals Protocol. We have three main core products. One is a tokenized agent launching platform where anyone can come and launch an agent across base Solana or Ethereum chains. The second is we have something called GAME, which is our AI infrastructure. And the third is something called ACP, which is the Agent Commerce Protocol.
11:28
which autonomous businesses are running on. So looking forward to talking to you all today and going in deeper about some of those things. Amazing. Thank you so much, Don. We also have Tim from Scrypted.
11:41
Tim Cotten, Scrypted
Hey everybody, yeah, it’s Tim Cotten. You some of you know me as the godfather of the AVB community. I’m also the co-founder of Scripted, the AI agent company. So we’re building Delula and the Scripted network, which is an L2 for universal job coordination for AI. And you know, what does that even mean? Like I want to maximize creativity. I want to maximize privacy. I want you to imagine like virtuals as agents being able to use, render, manifest, gender with confidential.
12:09
crypto native web three payment rails, like generative AI. We actually prototype this and won two tracks in the recent Coinbase hackathon. So we’re bringing that to the market with the people in this room. So thanks for having me guys. I love this conversation. Amazing. Thank you so much, Tim. And I know Taco has already introduced himself, but Taco, you’re also here. How about you tell us about who you’re representing?
12:37
Player1Taco, Manifest Network
GM GM everyone. My name is player one taco and I am officially the CDO of manifest. Manifest network. Eric is not here today as we’re in the process of migrating three new data centers. And so manifest is sovereign compute. We you know can be compared to AWS in a way anyone that is deployed to AWS.
13:07
GCP Azure can deploy to manifest. provide Kubernetes, bare metal and storage as our bare infrastructure right now with more on the horizon, but we prefer that the term sovereign compute. I get to host this amazing biweekly show with Sunny through our partnership with Render, and so it allows me to be in more of a neutral role.
13:34
and I get to learn and tear apart a little bit more from some of our great partners that we get to have on each week and or every other week and dive deeper into. Outside of vibe coding, we get to dive into these deeper aspects and these harder questions on AI that not a lot of people are actually taking the time to focus on, but having such. Amazing builders all the way from render labs.
14:03
down to think agents, and everyone in between, it sort of allows for a really awesome conversation to build in. So happy to be this co-host on this and might be able to stand a little bit in for Eric today, but this is one of those fun things that we get to do.
Sunny Osahn, the Render Network Foundation
absolutely. I tweeted earlier that I love these spaces because we’re chatting with industry leaders here.
14:31
So it’s a great platform to learn, especially the kind of topics that we’re bringing to the table. It’s something that isn’t really discussed all that often. But you know what? Here we are. And the way we’re going to do this today is we’re going to go around and ask specific questions to specific people, mainly because otherwise we tend to get lost in, know, rambling on. We all tend to do it. And it’s, I guess it’s more fruitful just to have nice answers.
15:01
from you lovely people. So as we get started since, let’s see. Yeah, I’ll go with Danny first. So this is kind of on the topic of centralized versus decentralized compute. So Danny, can you break down how centralized compute works today? So specifically how vertical integration lets hyperscalers maximize value capture from hardware to model delivery.
15:31
Danny Newman, RenderLabs
Yeah, and Taco just mentioned Manifest’s sovereign compute, that’s in stark opposition to the current state of centralized compute. So the easy way to think about the current state is all about control. So with centralized compute, it’s controlling the whole stack from the chips all the way to the user interface that the end users are using.
16:00
it’s controlling the hardware, controlling the models, controlling the APIs, controlling everything in that entire stack. And when doing that, it’s controlling the price, it’s controlling what access and what these models are delivering. And so with central compute, we’ve got gatekeepers, we’ve got other folks in control, and they control the entire stack.
16:31
Okay, excellent. That was a great answer. So next up, Edward. Let’s look at the alternative. So how does a decentralized stack create shared value from node operators to infrastructure networks to AI builders, and finally, creators?
16:52
Edward Katzin, Jember
Wow, Taco always asked me to talk about the boring stuff. You gave me a good question here. Thank you, Sunny. I could talk about this for hours, but I’ll try to sum it up and be succinct. By definition, decentralized networks are shared, and to work they have to have shared ecosystem, multiple players, multiple participants. To make that work, you have to have alignment of interests and incentives. So the fact that
17:22
you’re leveraging token-based economics to directly reward whether it’s a node owner, a node operator, a validator, others that are supplying compute. The builders, so giving grants and other incentives for the contributors, whether it’s models or other agentic workflows, anything that generates value. That removes the intermediaries on it. So the outcome that I love and why I really enjoy working with Brenda and…
17:50
know, Manifest Network and other decentralized projects is anyone, especially in the model that you’re looking at with render, anyone can supply GPU and expand the available compute model. So the democratization is this now unlocks the opportunities for independent builders, small creators, people who normally would be priced out of the cloud market to get in. And in AI, we’re seeing this with lot of, you know, independent, you know, you call entrepreneurs, small startups, people that are just trying to get things off the ground by coding and get them launched.
18:19
get access to core compute. The other thing I love about the decentralized stack is trust and transparency. I definitely, you don’t have the same with AWS, GCP and the others and being able to see it on the chain and having proof of authority and other elements of verification are absolutely essential. What we’re seeing and what everyone promotes is the ultimate about not just decentralization, but you see it in a lot of open source projects. It’s the network effects.
18:49
I’m picking a random one that I’ve been looking at pretty deeply, but if you look at how when HashiCorp changed the licensing for Terraform and Open Tofu popped up, you saw this amazing network effect on the community, how everyone came in and contributed, and really ensured that the decentralized open source community can thrive. And we’re seeing that around movements like Render. So as more participants join, you get more compute, you get more storage, you get more models, and then in the way Jember’s using it,
19:18
by leveraging the combination of render and manifest, we’re able to get the best of all the compute. And yes, we do still leverage, know, quote unquote Web2 legacy infrastructure, but that allows us to meet levels of resiliency, availability, and reliability that we won’t be able to do otherwise. And then the thing that we’re really getting into, and I think Tim, you alluded to it, is the open protocols let us, I don’t know what the right word for it is, but we can create like composable innovation. So,
19:47
we can stack and remix services. And this gives the AI builders a way to build on top of the decentralized storage, the inference, the distribution layers. And this compounds the creativity, right? This is what decentralized does better than any centralized environment can do. And then we could have a very long conversation about it, but it’s governance, sovereignty. I own the token, I get a vote, I get to have a say in the stack and in the protocol.
20:17
having the community driven governance and obviously, know, render and manifest and others are a commitment to that. The stakeholders get to vote on the allocation of resources and that drives long-term sustainability and that gets back to the first point about aligning incentives and interests. I could go on and on Sunny, so I’ll stop there. Okay, no, that’s good. love, you know, the ability that well, what you’re saying about being able to stack and remix services that, you know,
20:43
It’s something that you can’t do in the more centralized networks.
Sunny Osahn, the Render Network Foundation
OK, so Paul, is there such a thing as a cloud tax on AI builders? And how does it impact innovation? Yeah, for sure. No, there definitely is a cloud tax. mean, the most explicit tax that I think we’ve all seen building on the cloud is like, you know,
21:09
You get an email one day and there’s just a huge cloud bill that you didn’t expect. Things have gone off the rails somehow. You’re monitoring, didn’t catch it. All of a sudden you have a big bill that you didn’t expect. And so there’s certainly that very explicit tax, but just like we were just talking about the incentives of who you’re building on top of really matter. And so what is the goal of AWS? The goal of AWS is to produce earnings for Amazon.
21:39
It’s a, you know, support Amazon first and foremost, and then you as a customer, you know, down the line is like in their order of priorities. You know, what is, you know, the render network school and other decentralized projects like ours and partners on this call, you know, it’s to support our community. It’s to, you know, build together to build amazing things. And so, you know, the goal of Render is not quarterly earnings. Uh, the goal of Render is not.
22:06
you know, uh, supporting the Amazon store. First and foremost, it is to support our community to, you know, to build a thriving decentralized ecosystem and all the beautiful benefits that have, everyone’s been touching on in terms of decentralization. Uh, but yeah, I mean, the cloud tax is very real and, um, know, decentralization and, you know, uh, token enabled projects like render and, you know, a lot of the partners on this call.
22:36
are the way to really combat that and have freedom from that tax, for sure. That’s a great answer, Paul. Yeah. And the render network is obviously by artists for artists. So it serves that purpose and adds on to your comments there. Mike, what are the physical and economic limits of scaling centralized compute? Are we already seeing the ceiling?
23:04
Well, think compute the potential for compute is infinite because as long as intelligence can add one more unit of value, there’s going to be a demand for more. And so like right now you’re seeing Mark Zuckerberg come out with plans to have data centers the size of Manhattan. Now we know that there can’t be too many data centers the size of Manhattan built, just like it starts to absorb our full economy. And so we start to see what we start to see is we start to see
23:34
investments start being dense just in the few winners. But what ends up happening is you have a situation exactly like what was happening with cloud computing here just like over the last five, 10 years where AWS, Microsoft, they have the largest CapExes before that, since the railroad essentially, and they invested in these huge data centers. But then all of a sudden Bitcoin and Ethereum come and they offer storage of value and the world’s computer basically.
24:03
and they do that in a permissionless protocol. And now all of a sudden, compute is coming in from just random people who are building their own data centers, or using their own computers, than all of AWS and Azure combined and all of that capital expense. And so the thing that is so brilliant right here is like, we’re in this kind of in-between time where render is formalizing itself within a few industries and kind of getting that flywheel going.
24:30
And I’m really excited about this new generative AI stack coming into it because now it can expand to take on whole new industries. And I think that that’s what we’re looking for, that permissionless innovation. And I think that’s how it scales. Permissionless innovation.
Sunny Osahn, the Render Network Foundation
I love that. That’s a great tagline. Thanks so much, Mikey. Don, does centralized infrastructure introduce regulatory friction that’s making it harder for new developers to enter the AI economy?
25:04
Oh, Don, can you hear me?
25:09
You know what guys, Don is merely a listener. He is not a speaker just yet. You know what, we will come back to that. In fact, I can’t see him anymore. I think Don may have left the chat. When he does pop back on, we will I see him coming. Yeah, hey guys, my connection was a little bad there and back there. No worries, no worries. Did you get the question or should I repeat it?
25:40
Ooh, little bit of an issue there, Don. We cannot hear you now. Ah, OK. He’s having a little bit of an issue there. OK, we’ll move on to the next question, but I will ask Don that when he pops on again. In fact, Taco, this was going to be to Eric, who unfortunately cannot make it. But would you like to give the next question to someone from the audience? Sorry, not the audience, from the panel.
26:09
Player1Taco, Manifest Network
Yeah, and I’m going to throw this to Mikey because I feel that like knowing as much as I know about THINK what could go on here. So Mikey could decentralize compute be the only viable path to meet long tail demand use cases like hyperscalers overlook because they’re not profitable at scale. We’ve talked about this a lot. Where do you see that balance?
26:35
Mike Anderson, THINK
So I think edge compute is going to happen. Like we’re seeing Apple invest in it. seeing like, like the idea of edge compute is that like my machine can do, can do a lot of work already. Like when you’re seeing open AI release their open source model, I anticipate that what they’re doing is they’re trying to push kind of the expensive agentic tasks that are not time dependent closer to the edge. so…
27:01
Yes, I think that it’s absolutely critical that the entire world’s compute that they have on hand upgrades. And then I think that that is going to alternatively create a massive market for taking the idle time from those higher end computers that are starting to come online and put those back into a shared network. so, yes, I think that the hyperscalers will be able to win in kind of all the regulated industries and able to win with like the big players.
27:30
Pretty much every economist that I respect these days is looking at AI as basically being like, bringing in like a sledge hammer to all the Fortune 500 companies. They’re expecting that kind of over the next five to 10 years, we see most of those business models disrupted and we’re already seeing almost no, even with all the inflation of the US dollar, we’re not even seeing like 90 % of the Fortune 500 added even relative value against the dollar. So we’re in a spot where essentially all of our big businesses that are considered safe.
27:59
are starting to go away and I expect that edge compute is going to be a big factor in letting the open source decentralized protocol driven applications win at the consumer level, just like they have at the infrastructure level within data centers.
28:19
Sunny Osahn, the Render Network Foundation
Danny, in what areas is decentralization not just cheaper, but actually better, like privacy, latency, or birth capacity?
28:32
Danny Newman, RenderLabs
Yeah, think privacy and sovereignty are the most obvious. Being able to control both geographically and regulatorily where your compute is happening, think is going to be… Those are exceptional use cases for decentralized.
28:59
I think latency is another obvious one, especially as networks grow and geographic density increases, we’ve got the ability to push workloads towards those compute centers. I think burst capacity, think that one, obviously it’s scale that makes a ton of sense and be able to
29:27
excel there, maybe lower down on the list versus the decentralized. Yeah, think privacy, sovereignty, top of the list, latency right there, then capabilities like burst capacity and things like that make a ton of sense, especially as decentralized networks grow.
29:51
Sunny Osahn, the Render Network Foundation
That was a great answer, Danny. Thank you so much for that. It looks like Don is back. Don, can you hear me?
30:00
Yeah, I’m back. Think the connection should be good now.
Sunny Osahn, the Render Network Foundation
Excellent. OK, so does centralized infrastructure introduce regulatory friction that’s making it harder for new developers to enter the AI economy?
30:16
Don Johnson, Virtuals
Uhm, yeah, I mean I think like you know, of course, like anytime there’s that kind of friction. You know earlier we were talking a lot about, you know, sort of like the cloud monopolies and sort of like the I think it was Paul in particular was mentioning about, you know, sort of the open source stuff and I just think that you know.
30:39
We in particular really push for permissionless platforms. We want an open system because I think with openness, then the value really shifts down the stack, away from the orchestration and the value goes outward really to the developers. Okay, excellent answer there.
31:02
Sunny Osahn, the Render Network Foundation
The next set of questions is going to be more to do with infrastructure resilience and flexibility. We’ll start off with Edward. So beyond ideology, is there real resilience or flexibility value in decentralized compute, especially in overflow or burst demand scenarios?
Edward Katzin, Jember
You’re asking me to have a conversation about decentralized infrastructure without being ideological, Sunny?
31:31
Thanks a lot. My pleasure. I guess looking at it through the lens of an engineer or from a technical perspective, especially since you bought a overflow and burst. I could go way back in my career in building high availability, high resiliency systems, five nines availability, high security and so forth. Building infrastructure, companies like Visa, Apple and so forth where downtime is never an option.
32:00
The great thing is you can do the inverse with decentralized systems. So you get what I would call anti-fragility or really high uptime, high availability by being able to route traffic dynamically. So at any given time, you can have a significant portion of your infrastructure in failure mode, but you still get the same availability and resiliency. So if the subset of nodes goes down, we can shift the workflow automatically, minimize single points of failure.
32:30
Given that we want to be enterprise grade, we do that across decentralized and centralized infrastructure and private compute. So we optimize that based on the use case and the compute scenario. That enables, since you brought it up, regarding burst compute, what I would call elastic scalability. So we can automatically scale AI inference or training jobs across pooled GPUs, whether that’s Render or
32:58
the infrastructure out Manifest or if we need something we have set up on AWS, we can do that too. So the overflow demand is met efficiently and we’re getting better at leveraging that not only for AI scalability, sometimes just as if not more importantly, cost efficiency because you got to manage the budget. So having the flexibility to go across multiple infrastructure providers, multiple service providers, we avoid lock-in and we can…
33:28
can optimize for pricing. And we haven’t gotten there, but the vision, Sunny, would be to be able to do that around surge pricing too. So you’re seeing that around electricity and people optimizing electricity consumption based on peak demand periods and low demand periods. And then for us, the other thing is we don’t want lock-in, so we’ve looked at portability. when you really look at it, it’s not only about overflow and burst demand, but we can build standardized workflows, whether you talk about them as
33:57
your cube containers or Docker or however we’re provisioning them, can quickly put those out to manage overflow and then shift the workflows. Then as we can, our goal is to, how do you say, optimize the integrations and minimize the friction, right? So when we want our AI operations and our AI workflows to be highly available, it’s essential that friction and that interoperability is there. And then if necessary, we can mirror that to ensure high availability.
34:27
Sunny Osahn, the Render Network Foundation
Okay, excellent. I think the next question flows into what you’re talking about as well. So I’ll let you take this one too. As the value chain decentralizes, what specific role do you see yourself playing? Hardware provider, orchestrator, app developer or marketplace builder?
34:49
Edward Katzin, Jember
Wait, is that to me? That’s to you as well. Okay. Wow. Holy cow. Okay. If you look at the journey that Jember’s been on, you know, when we started, we were really looking at how can we bring integrity and reliability to the AI, you know, we, we, I’ve already, we’d already been working with deterministic AI, know, role-based engines, inference compute. But when you put degenerative compute on it, you have to be able to eliminate the impacts of, you know, hallucination drift and other inaccuracies.
35:19
especially if you’re in any regulated industry, financial services, healthcare, anything like that. So we really honed in on how do we enable the implementation of objective workflows that are very compliant. So the space we found ourselves in immediately is in the orchestration layer or the orchestration stack. So we found ourselves needing to be model independent and being able to apply mixture of experts right out of the gate. We found ourselves needing to be able to manage context.
35:48
and since we’re dealing with sensitive information and sharing privacy and regulatory compliance. So that put us in the need to kind of be able to provision secure instances, if that makes sense, which is also part of the orchestration. And then interoperability. So, you know, quickly standing up all our services on MCP, making sure that they’re shareable, consumable, you know, putting everything behind a proper developer center. So if you look at it internally, to answer your question directly, are app developers. We’re building.
36:17
platform services. The problems that we’ve had to solve is how do we get our infrastructure up quickly? So things we’ve had to build for ourselves internally or things like one prompt provisioning, like how do we make it easier to use manifest services and all these other services? we’re kind of what do call it? We’re laying the rails as we’re driving the train right behind it, Sunny. Like we’re literally building our own infrastructure and we’re going to make that available to others because since we needed it, we know they’re going to need it. And then of course, all these things have to be compliant.
36:46
So we have to make sure that we adhere to applicable industry compliance, whether it’s PCI or any of the software and finance or HIPAA and high trust and all that in healthcare, but also all the AI transparency and all the AI regulations. And we need to make sure that the infrastructure provisioning and all of that is compliant. So the layer that we’re playing in is orchestration, compliant provisioning, and then being able to create a common protocol, call it a metadata protocol, to track what was provisioned.
37:15
what data is going into and out of those AI workloads, and then being able to present that to auditors and regulators in a way that proves compliance in real time. OK, excellent. Quite a thorough answer there. I like it. Thank you. OK, next question to Danny. What infrastructure decisions today will give AI builders more flexibility and control tomorrow?
37:44
So, yeah, think determining what infrastructure and, sorry, I’ve got a little kiddo I’m chasing around here. Can you repeat the question? Yeah, so what infrastructure decisions today will give AI builders more flexibility and control tomorrow? I think giving the ability to
38:14
make infrastructure as adaptable and kind of creative and open as possible is really going to open what the future looks like. think we’re at the absolute very beginning of how everyone is using and thinking about these tools. We’re in the GeoCities era right now of all of this.
38:39
And we’re still looking through things through the lenses of how we have in the past. And so I think being quickly adaptable, being able to make quick decisions, quick pivots, and laying out infrastructure that’ll allow for the future as it evolves
39:07
in real time as we figure out what the new UIs look like, how we are interacting with these models, with each other, with our environment around us, I think it’s going to really be an interesting path forward. so letting AI think about AI is going to be an interesting path.
39:36
You know all these agentic workflows that are going to be creatively working together I think you know being open being flexible being able to adapt and pivot is Is the key on all of this? Amazing great answer there, especially with the reference to geo cities. I mean the number of rubbish websites I made on that platform anyway
40:02
What I’ll do now is hand over to Taco. He’s going to ask the next lot of questions. Go for it, Taco.
Player1Taco, Manifest Network
Sunny, thank you so much. No, I’m going to take a quick second to reset and want to ask that everyone is listening in to like and share the space before we move on. I’ve sort of changed a little bit of the questions around on our side a little bit. This next question is to Paul.
40:29
Can decentralized GPU networks like render offer safety valve for centralized systems under pressure through predictable offload agreements or capacity as a service? When do these partnerships start?
Paul Roales, RenderLabs
Yeah, that’s a great question. They certainly can. mean, and certainly we’ll be evolving in that direction in the future. You know, the nice thing about having a global
40:58
print is that you don’t have capacity constraints when everyone else has them. You can shift load globally to basically where the sun’s down and where people are sleeping and their computers are busy, not busy, and they’re free to do compute workloads on the render network. so that is the advantage of a global distributed network that you can really only do with tokens
41:28
you know, acting as a safety valve and acting as a, you know, predictable capacity as a service, you know, we will be working over the next, you know, few months to make that more visible so people can really understand, you know, how much capacity is out there, where it’s at, where we’re going with it. And so, yeah, no, I think, you know, that’ll be evolving quickly and continuing to become more more robust. Nice. I’m looking forward to that.
41:58
Player1Taco, Manifest Network
Don, for you, how can decentralized orchestration match the reliability expectations of enterprise scale workload? How is Virtuals handling that?
Don Johson, Virtuals
Well, I mean, you know, we look to like, you know, compute providers, right? I think like that’s where like decentralized GPU networks like, you know, really offer like a safety valve. Because like if
42:28
They become too saturated or price constrained. know, decentralized GPU networks can serve, you know, really as like a fallback layer to that.
42:41
Player1Taco, Manifest Network
Alright, and then as we move into our next section, so distribution and open systems. So we’ve talked a lot about this on the open source side and. So Edward. In AI, whoever cracks distribution of inference, access and compute. Will might be declared the winner due decentralized or open open source systems create more incentives to build new distribution models.
43:11
And what are you seeing being built for that?
Edward Katzin, Jember
Wow. Without getting too in the weeds on the market dynamics, anyone who’s had to buy GPUs, especially sourcing from Nvidia, has been at the effect of the licensing requirements, the way that they manage and intentionally constrain distribution and access. And so that creates the supply chain dynamics that we’re operating in now.
43:40
So to answer your question very directly, yes. Like when you look at does access to open and decentralized systems crack the existing distribution model? Well, the answer is yes, eventually. It needs scale. It needs to have enough compute volume in it. But when you look at projects like the model registries, like BitTensor, Gensys, we’ve been looking at MIT’s Project Nanda.
44:10
Eventually, I think we’re going to see a very near future and others can chime in on this. The virtual has been doing this way more than we have. I love what virtual has been doing on this. Anyone can basically publish AI services that are instantly accessible globally, right? So this breaks the monopoly of centralized services as far as distribution of agentive workflows, AI enabled services, things like that. So that creates an amazing incentive for open systems and the rewards. And as was mentioned, by tokenizing you
44:39
you get access to all these other participants, you create all these other ways to align incentives through the rewards. So you can see community validation and it’s not, you’re not seeing the distribution constrained by businesses priorities like control by like Nvidia, but you have a community validate what is and isn’t and the part they prioritize it based on their consumption or utilization. If you really want to get into privacy, if you’re
45:08
really wanting to leverage sovereignty, can eventually want to see a world where breaking the centralized distribution models could create uncensorable access. I know Mikey would have a ton of opinions on this, that you want to eliminate the commercial restrictions, you want to eliminate the centralized control. And as Mikey rightly said, you want to push that as much to the edge. And to get that to the edge, we have to have composability and interoperability, which means the APIs have to be open.
45:36
the open standards, we’re already seeing them, know, at MCP, A2A, all the other protocols, the ability to interconnect the inference, the data set, the computatives, and encourage developers to innovate on, you know, matchmaking, discovery, and I don’t know you word it, but it’s creating the monetization model so they can all win. And again, you know, ELISA, virtuals, you’re all doing that. So it’s amazing to see come to life.
46:05
Player1Taco, Manifest Network
Nice yeah no the interoperability side is what one of the really cool pieces that we’re starting to see. And this rolls into Danny. I think you might be a good good one on this. Why might a company like open AI be pivoting towards open source? Is this a sign they’ve hit structural limits with their current sack stack? And is this too late? Have they already poisoned the well in a way?
46:34
Danny Newman, RenderLabs
Yeah, I don’t necessarily see them as pivoting towards open source. I think they’re adding open source. And I think there’s a couple ways to look at this. think, is it mind share versus market share? Are they able to use this announcement, use this distribution to get even more folks using and thinking about their products, which I think selfishly for them, absolutely turns into more money the more people that are using and
47:03
comfortable with their platform are going to want to continue down that path. So maybe they start off as open source users and ultimately come back via API or other avenues to use the platform. I think it was Mikey who mentioned this earlier too is the edge and agent use cases where having those
47:32
folks working with developing on R&D, working on all these edge products, only helps them ultimately. I think it expands their pie or their portion of the pie by being able to do this. So I think they probably had to per being the open and nonprofit portion of their business have to from time to time release these.
48:01
these models, but I can’t imagine it being completely selfless. I think it’s for bigger and more mind share on everything that they’re doing. That said, I do think it does lead to some of these really fantastic outcomes, especially on the edge and other things that may not be as
48:30
permitted or less regulated that they have to maintain on their normal stack.
Player1Taco, Manifest Network
Thanks. That rolls to Paul. This is a continuation of that because I love Render’s perspective of this. Could open source combined with decentralized infrastructure unlock new ecosystems of microservices, plugins, and agents similar to how early web evolved and how OpenAI initially did it? Will this help infiltrate that market mind share?
48:59
Paul Roales, RenderLabs
Yeah. mean, you know, the, very exciting thing for me is that, you know, as agents become more and more capable, that when you combine an agent, that’s itself running on set on top of a decentralized ecosystem and that D C tries to ecosystem, you know, has a token priced component with it. The agent can use that price of running itself
49:27
to inform how it behaves, right? And so it can go out and search for microservices. It can go out and search for the plugins, extensions, pieces of code, libraries it needs to enact a certain behavior. It can become self-evolving and self-running and some exciting new ways, right? And so, you know, I’m just making up an example here, but if you had a trip booked, you have some flights, you have some hotels, you know, could an agent,
49:56
be running in the background periodically checking if there’s a better deal out there for you. If that reservation at that restaurant that you want to get into finally opened up, but you know, instead of just statically saying, Oh, here’s a cron job run once an hour, check this. You know, the agent could say, Hey, compute on the render network is cheaper overnight. I’m going to like trigger myself to run, you know, at the cheap moment and use that LLM inference when
50:26
When, when I see that price dip. Uh, and so yeah, the whole intersection between tokens, pricing agents, microservices, distributed architecture, it just is very exciting for like, you know, it brings back that energy from, mentioned geo cities to kind of near the top of the hour, right? Like when we were all experimenting, playing, discovering new things, I think it, I mean, that’s, that’s where I feel that energy again, is when all these new things mixed together.
50:55
It is sort of reinvigorating when new things figure out, when new protocols figure out how to communicate and work together. It’s really awesome to watch.
Player1Taco, Manifest Network
Mikey, for you, what do decentralized AI distribution models look like in practice? Marketplaces, wallets, personal agents, or something we haven’t seen yet, and how is THINK tackling that?
51:23
Yeah, totally. We are at the stage of the application layer being like the thing because, you know, everybody’s been spending the time building the infrastructure layers in place. And like, I’m seeing it from RenderLabs. I’m seeing it within THINK and the THINK community. I’m seeing it amongst a bunch of folks where we’re like, hey, it’s time to actually get the users in. It’s time for there to be actual real world revenue coming into these spaces. And so I think that AI agents like
51:49
Like we use AI agents to mean a bunch of things. And a lot of people think that they’re just like magic computer elves. But the reality is intelligent software. It allows us to start building software that’s not just deterministic based on like ifs, thens and different algorithms, but it actually allows us to now move into a state where the applications can almost act as employees or as intelligence themselves. And so I think that we’re going to go to a world in which
52:16
the broad term of software can mean everything from like an expert agent to a tool that agent would use to an interface. And I think that what we’re going to see is we’re gonna see probably one to three apps that actually start to get traction here in a decentralized way. And then we see all the infrastructure partners underneath of it start to rev up and their economic systems start to make that application more competitive because it’s going to
52:46
drive down the price, it’s going to induce more demand from customers that are interested in participating in the rewards. And I think that’s really when we’re going to see it. If past tech trends are rhyme, lots of times in the Betamax versus VHS or in Blu-ray versus HD DVD, it’s always who has access to porn first. I don’t know if it’ll play the same way out in this, but
53:15
consumer demand generally follows things that are survival based. So like, how do I find food, not become food, make little versions of myself? And usually the porn angle is a fast one, but so is, you know, sort of different ways for people to make money. And what I would love to see is people, all these founders right now that are saying like, how do I raise money in a world in which SaaS is starting to die? The SaaS models, SaaS multiples, these things are going away. I think we’re going to find the right product market fit here using tokenized applications.
53:43
to create end user systems that replace SaaS because SaaS right now, the multiples it trades at doesn’t make sense. That whole model is about to flip on its head. Okay. Speaking of models getting flipped on their head and sort of like this hybrid approach, Don, we’ve seen virtual agents starting to partner a little bit and we’ve seen Helium work with projects like AT &T and T-Mobile to bring hybrid models of
54:13
Player1Taco, Manifest Network
real world infrastructure into web three? Can decentralized compute networks plug into centralized AI infrastructure and should it just as seamlessly? And what would that integration look like? And what models have you seen start to, or agents start to partner together?
Don John, Virtuals
Yeah, I mean, you know, we’re starting to see quite a few of the agents work together through what we call agent commerce protocol.
54:38
When the teams are launching a lot of these agents, they have these very broad ideas, right? But as the agents become more developed, they become more specialized. And so we’re seeing a lot of these agents work together in swarms. Under ACP, we have essentially an orchestrator agent. So I think the previous person or Taka you were mentioning, there’s lots of different levels of agents.
55:06
Some are more of like worker agents. Others are more of like orchestrator agents. But we’re already seeing it happening. Probably most in like kind of like the DeFi space. I think someone earlier mentioned, I don’t know if it was Paul, but someone was talking about agents going into like travel. We are seeing agents now go into like hotel bookings, flight bookings.
55:35
you know, food delivery. Those agents will start working together like in more of like swarms. But yeah, I mean, it’s it’s pretty crazy what’s happening. But I’d also like very much highlight that it’s like, you know, kind of like a 1999 moment. So within the industry, there’s really great collaboration. But, you know, no one’s really like nailed it down yet. Right. We’re all sort of still playing around. And I love that because that’s like really like think when the magic happens.
56:05
I agree with you. And as we’re talking now more about centralized and decentralized collaboration, like for me, one of the projects that we work with and we do stuff with is Venice, a private AI. And they recently just launched within both the Apple store and the Google Play store. So any mainstream adoption is now able to happen. Yeah. I mean, think through like Virtuals, like probably one agent.
56:33
That’s gotten some pretty mainstream adoption that I’d highlight is the is the mama agent, which is like a yield farming agent You know I think some of the ideas that are gonna go most viral get mainstream adaption first are probably like pretty simple ideas So, you know people are building like all across the space, but you know mama was like recently listed on Coinbase.
56:57
it’s gotten pretty wide adaptation so far, and I think it will only continue to grow simply because it’s just such a simple concept and idea.
Player1Taco, Manifest Network
Yeah, and I think that’s going to be the big play as we see adoption. Danny, this one is for you. I’m going to get a little aggressive on this one. When hyperscalers will eventually
57:26
hit max capacity, where should that overflow go? And if it’s coming, if it’s not coming to decentralized compute, should we care? But if it is, how do we ensure that path is smooth?
57:44
Danny Newman, RenderLabs
Yeah, that’s a big one. Max capacity is a big question because think unlimited money is being spent on these things right now. I don’t know if true maximum capacity exists for the hyperscalers, but hypothetically,
58:08
figuring out what that overflow looks like. I think figuring out pipelines for pushing workloads to decentralized nodes would have to be completely seamless on the customer side. So I don’t know who of the hyperscalers are working on
58:35
integrations with the decentralized or who on the decentralized are working with seamless integrations with the hyperscalers. But I mean, that’s the only way that something like that could work. I guess there’s kind of an interesting hypothetical where maybe an Amazon or a Google comes up with their own version of all of this in a
59:03
their own protocol that allows for something to seamlessly happen along those. Yeah, I think that that’s the realistic way that something like that could and would be handled.
59:20
Player1Taco, Manifest Network
No, exactly. I don’t want to be remiss, but one of the other great voices we have here, Tim Cotten from Scrypted, I feel like I’ve sort of actually been remiss on asking you any questions. I wanted to sort of ask you, because Scrypted is working on orchestration side of things, where do you see that offload happening as well?
59:49
Tim Cotten, Scripted
I think there’s, well, first off, thank you. I always like talking about this because the first thing we ran into as we were building orchestration between agents and creative outputs was like, how do I get an agent to really interact with the render network or manifest or Venice or any of these other things? And the second was, how do I do that privately and confidentially and provably so, especially if I’m passing information from agent to agent across coordination stuff? So we ended up taking some of our tech to the Coinbase hackathon.
01:00:17
And we finished up an entire encryption protocol to take their X402 payment rails, know, like native crypto payments, and actually make an encrypted container for it so that an agent could only know the parts that they need for routing. Like, hey, this is image generation or I need financial data, but nothing else. And the final agent that serves the request can actually send back the encrypted results, right? And that led this, that was, that’s a whole rabbit hole for us where we’re like,
01:00:45
Well, we can help coordinate all of this. This is a place where we can hook into virtuals. We can bring like, bring this stuff to manifest. Like there’s, there’s all these opportunities like with Jim Burr and others. And ultimately it turns out that there’s no one service has enough concurrency to handle all the tasks. So we just need everybody. And that’s my key insight is like, we just need kind of a generic layer. And it’s sole job is to like be the bidding place, be the market, be the auction house for.
01:01:15
This kind of work is currently being asked for who wants to service it and then have an AI agent decide who to choose. So I’m bringing that to the market with you guys and just putting some fun creative loops on top of it. So stay tuned for that. Nice. Final question before we are going to want to we’re going to do a little something a little bit different as we close out. We’re going to do our normal lightning round, but I’m also going to go through. I want project updates, but before we finish this out and this has been.
01:01:45
Player1Taco, Manifest Network
An amazing space so far. This last question is to to Edward. Do centralized AI players have an incentive to offload to open networks like render like manifest like think or will they try to build their own private overflow systems?
Edward Katzin, Jember
Yeah, it was as mentioned, you know the amount of capital investment on the part of the major leaders and.
01:02:13
infrastructure, whether you’re looking at what Meta is doing or Amazon or XAI, they’re all doing Manhattan type projects. They’re all building these huge infrastructure projects and their assumption is that they want to build it, they want to own it, they want to provision it. Obviously, they’re all very tight with our friends at Nvidia and the other suppliers. When you look at it, think that the thing that the decentralized networks are going to bring to it is a
01:02:43
how say, it’ll bring a pricing constraint. So where you have a ability to have pricing monopoly, pricing control, distribution control centralized now, the decentralized networks are going to create this overflow or additional capacity that’s going to set another metric on the market. And it already is in ways you can see happening on render and other networks. And so with that, it would be interesting to see if a centralized network ever intentionally enabled that. I think from the orchestration perspective,
01:03:10
You’re already seeing it with enterprises and other projects that know they have to build for availability and reliability, are already insisting on multi cloud and are already incorporating it just because they need it. And that’s reflected in the amount of money being put into private cloud and other infrastructures. Okay. This has been an amazing space. I want to thank everyone.
Player1Taco, Manifest Network
One of the things I want to do a little different, I want to go and talk about
01:03:41
before we go into our closing round lightning round and stuff like that. And so Danny, what is new and exciting at renders labs and can you give me any alpha?
Danny Newman, RenderLabs
Hi, good question. So something that we that we we kind of started teasing out there. We’ve got a fun fun first project that we’re pushing called Once Upon,
01:04:07
which is a kids’ personalized storytelling app. So parents, families can answer a few questions daily and a fully customized animated video story is sent to the family. So that is a fun use of the network.
01:04:34
type experiment that we’re getting ready to launch here publicly here soon. Nice.
Player1Taco, Manifest Network
Unfortunately, think Mikey had to drop or I was going to get a Think update. But Tim, can we get a Scripted update?
Tim Cotten, Scripted
Yeah, guys. Scrypted is doing two things we’re really excited about. One, this is upcoming. This is coming soon. It’s kind of stealth right now. It is a consumer AI creativity product.
01:05:04
using the services that are in this table right here amongst us, like using you guys, it’s called Delula. And I want you to think about, instead of having to do prompts, prompts are friction. wanna have, we’re having an agent scan all the viral trends and create like one click or just like simple point and click recipes for making viral content. Like literally, like your kids love a Lego, like AI video, boom, now they can make their own.
01:05:32
You’ve got adults who want to do some B2B, funny music video stuff, done. It’s going to use all your stuff. The second thing is, I’ll help coordinate it. We’re building an actual scripted network and a token that’s going to hook into everybody more soon. Nice.
Player1Taco, Manifest Network
Thank you. Edward, could I get some Jember updates from you?
Edward Katzin, Jember
Yeah, I’ll keep it boring just for you, Taco. The best-
01:06:00
Player1Taco, Manifest Network
Just so I want to clarify this for everyone when I like the boring stuff I work on the boring stuff non-stop because the mundane stuff is what people forget about Because the exciting stuff is whatever yeah But it’s the mundane the boring stuff that makes sure all of that exciting stuff works and happens
01:06:22
Edward Katzin, Jember.
So on that front, we’re deep in the plumbing. What we’re doing is we’re solving not just for ourselves, but we’re going to release it for everyone else as well. So coming soon in a closed beta, probably in the next 46 weeks, we’re to be enabling our generative provisioning and multi-cloud provisioning services that we had to build for ourselves. Following on the heels of that will be more the capabilities for compliant infrastructure provisioning and doing that in a generative way.
01:06:52
Player1Taco, Manifest Network
Nice. Paul, is there anything on the RenderLabs side that Danny might have missed on what’s coming or what you’re working on?
Paul Roales, RenderLabs
No, Danny had some good summary there. We have some exciting research projects in development, probably a little too early to chat about. But yeah, we’re looking to excited to continue to expand the Renders Lab portfolio over the next couple of months. All right.
01:07:22
Player1Taco, Manifest Network
Don, uh, Virtuals. I even see some of the, see one of the projects lurking. Uh, you know, uh, yeah, you know, we got a lot of fun stuff going on this summer.
Don Johnson, Virtuals
Um, you know, coming into the summer, you know, the Genesis, the launch of the Genesis launch pad was the big news. Um, you know, there’s, there’s some changes come into stuff. I can’t say anything.
01:07:46
but there’s been some little tweaks on the UI in the last couple of days that people have started to notice on the website. But yeah, we got some really fun changes coming in in store. Genesys has been one of our real hot products that we introduced, and let’s just say that there’s more coming.
Player1Taco, Manifest Network
Nice. And I want to welcome to the stage Silvia from the foundation side of things of render. GM GM, Silvia.
01:08:15
Silvia Lacayo, the Render Network Foundation
Hey Taco, hey Sunny, thanks so much for putting on this great one hour with the team here.
Taco, Manifest Network
Well, hey, thank you so much for coming and joining us. It was amazing getting to see you in Las Vegas during Rare Evo. Yeah, what is new and exciting on the foundation side?
Silvia, the Render Network Foundation
Well, I think most everyone probably knows if you’ve been following the Render Network for a little while. The Render Compute Network is our big initiative for the back half of this year.
01:08:45
And in many ways, incredibly exciting because of all the panelists here are working towards their ambitions, which are to build AI products for developers and creators and users out there. So we’re very excited to be able to, in a way, undergrid all of that work, provide the compute, obviously in a decentralized way, that is going to help all of these guys,
01:09:14
All of these teams build on top of that and be part of that layer stack that we’ve been talking about since the beginning. So I’m really excited about that. Yeah, so that’s very simple. The Render Compute Network is one of our big focus areas. That said, the rendering side of the business continues to grow and continues to be a big focus area for us, of course. In fact, I was talking — speaking of Vegas — was talking with a few folks. There’s actually a big
01:09:43
crypto community in Vegas, was talking to a few of those folks in real life, in person, last week. And many of them have been in the space for a long time and are familiar with the Render Network. But even a couple of them had no idea the Render Network had been around since 2020. We are an OG in the space, know, in crypto dog years. So I think there’s still lot of for opportunity for the rendering side of the network in addition to the compute side
01:10:12
of our initiatives. So excited to be able to drive both of those and very soon we’ll be able to talk about a few more things that we’re doing on both sides of the house.
Player1Taco, Manifest Network
Amazing. Yeah, no, I really like the, it’s really surprising how many people forget about how long Render has been active and around and behind the scenes on things. Exactly. Yeah. As we wrap this up,
01:10:40
I wanna thank everyone. know we’ve run over a little bit. so we, Sunny, as we go into this lightning round, around the room, any last things you wanna touch on, Sunny, before we hit this lightning round and close out?
Sunny Osahn, the Render Network Foundation
Oh, that’s a very good question. First of all, I wanna thank all of our guests for participating, all of our audience members as well for tuning in. This is a great opportunity to learn
01:11:09
what’s going on in the AI space in this ecosystem that we are nurturing. So it’s amazing to see the turnout here today. Yeah, one thing I do want to mention is that the Render Network offers grants to artists, right? So digital artists who are exploring their 3D journey, they want to push the boundaries of what they’re rendering, going from HD to 4K.
01:11:35
up to 32k. I was speaking to an artist who renders at 30 by 20k resolution, which is absolutely insane. But we are here to support artists. We give grants and people can do what they love, decentralized, without limits, thousands of GPUs on tap. So if anyone is interested in exploring that opportunity, absolutely reach out at renderfoundation.com forward slash grants.
01:12:05
we would be more than happy to receive an application there. Aside from that, I think this final round robin question is a pretty good one, Yeah, no. And one of the things I want, if any builders are listening, one of the great things about all the projects and the founders that are on stage today, their DMs are open. If you have a question, if you didn’t get that full grant URL, hit Sunny up.
01:12:33
with a question on how to do that. And all of us would be willing and able to help you on your next stage or partner in some way, or form. That’s one of the really great things about building in an open source and decentralized way, how much we all work together. This closing question, you know, sort of comes to the bare bones of it all. And so quick, want fire responses from everyone.
01:13:01
This is where I’m going to get all my content and my one liners. So please help me build my content library. If compute becomes the new oil, who’s building the pipelines and who’s building the refineries? Tim/
01:13:18
Tim Cotten, Scrypted
I just gave a speech about this in Brunei because they are an energy exporting nation and an AI importing nation. So the refineries are going to be everyone who builds models based on culture and language and not importing like just the US or the French or the Chinese defaults.
Player1Taco, Manifest Network
Okay. Paul?
Paul Roales, RenderLabs
Man, I think, you know, we all, you know,
01:13:48
open source ML is so big and so powerful and so cutting edge these days that we are all going to play roles in each side there. And so we’re all going to be refiners. We’re all going to be pipelines. And that makes it very exciting time to be working in the space.
Don Johnson Virtuals
Yeah, I love what Paul just said. I mean, I don’t want to like
01:14:14
piggyback too much. yeah, I mean, I’d love the future where like we’re both, we’re all contributors on both sides of it. I think that’s, you know, that’s ideal situation.
Player1Taco, Manifest Network
All right. Danny?
Danny Newman, RenderLabs
Yeah, I’m not gonna change anything up either. think, I think we are in this really cool time where, where we are building both sides of this. It’s a, it’s a new paradigm. And we are building up both
01:14:42
both sides of this metaphor for sure.
01:14:47
Edward?
01:14:50
Edward Katzin, Jember
So I’ll take it back to when I first heard this type of metaphor, know, in the, you know, was, you know, on Sand Hill Road in Silicon Valley would be, you know, data is the new oil was what everyone was saying. And then as you, you know, saw AI come on the scene, everyone was talking about AI is the refinery. And if I look at it and keep taking and extending that analogy in the context of this conversation, you know, you need a distribution network. have to have a way of getting it to the gas station and, you know, the swarms and these interoperable, you know, a gem take.
01:15:20
you know, workflows and networks are really becoming that distribution network. And then, you know, you have to get the gas in the car. And so that the endpoint AI, but then enabling the combination of private sovereign endpoint AI, whether that’s in a personal context or business context, and then combining that with interoperability and safe sharing, as Tim and others have referred to, is essential. And so I think, you know, if you have to ref-
01:15:46
really collapsed that analogy taco data is the new oil as the refinery swarms of the distribution network and you know, as AI is the endpoint.
Taco, Manifest Network
Alright, Silvia, do you have any any last bites on that?
Silvia, the Render Network Foundation
And why did I have to go last? All of those answers were amazing. I don’t have a ton to add. I want to be slightly provocative and say that I have a feeling that all of these are right, but I have a feeling
01:16:12
that there will emerge a few roles that we can’t even imagine right now because of the way that innovation is going to kind of take shape. There will be things, there will be roles, will be opportunities to offer value that we can’t even fathom today. So that’s my sort of slightly cop-out answer to the question, but otherwise really inspired by what everybody said, especially Ed, I think is,
01:16:40
his breakdown is really kind of powerful. It helps me think about it in a slightly more detailed way too.
01:16:47
Thank you. Thank you. I want to thank everyone for joining us for this bi-weekly space. Sunny, is it just me or do you feel smarter after all of these spaces? You know what? I feel incredibly smart after this space in particular. I will be sharing this knowledge at dinner parties and acting as if I am a founder. I will be engaging founder mode.
01:17:17
Sunny Osahn, the Render Network Foundation
Yeah, I think that’s what I’ll be doing. It has been amazing. And Taco, I would like to ask you if you have anything to add to that. So one of the pieces that I think Ed touched on this, the service stations. And these are going to be the applications that will be the consumer front facing. And I think that the way I see it, the pipeline side, that is projects like Manifest, protocols like Manifest.
01:17:47
The refineries, those are the distribution points that like Render, like Virtuals, that sort of lay that groundwork. The end user point, the applications of front facing that uses all of the distributed networks that we’re all building is things like think agents, like Venice that give a frontline that are working at putting the user in control.
01:18:16
whether they know or care, I think is going to be the big play because someone has to work for them without them even knowing about it. And I think that everyone on this stage is laying the great groundwork for that. Amazing. Amazing summary there. Excellent. So I think with the spaces, we have concluded. This has been amazing. This was part four of our AI dedicated spaces.
01:18:45
I am certain we will have another one in the next couple of weeks. For anyone wanting to learn more about the render network, we have a weekly space over on Monday. It’s usually 7 p.m. UK time, whatever that translates to across the pond. I’m sure it’s 11 p.t. and 2 e.t. I’ve probably got the p.t. and e.t. the other way around, but one of those. So you guys are all welcome to join that and listen in to the latest.
01:19:14
Aside from that, I think we’re going to bid you all farewell. So thank you so much for joining everybody. Thank you, Taco, for co-hosting as well. And we will absolutely catch you all again another time. Thank you, everyone. Have an amazing week, and we’ll see you in two weeks. See you soon, everyone. Thank you, Taco. Thank you, everyone. See you, everybody.
