Cloud Decentralization

“The idea of supercloud was just getting formulated — I think Cornell wrote a paper in 2015 — and we really liked the idea of supercloud, that you can decouple the control plane from the execution plane or the resource plane, and reduce the cloud providers, both public as well as private, to resource providers and not give away control.” — 00:03:36

“Every time we launch, we’re very, very, very cognizant about the value proposition and how we differentiate ourselves… So we believe we’ll achieve parity in the next two years. And after that, the phase we like to call is really innovation beyond imagination.” — 00:17:20

“Akash five years down the line and be decentralized, gives Akash a unique angle that traditional companies doesn’t, which is permanence. Like permanency is not possible with the corporation, only possible with a protocol… with permanency, you get time that you don’t get with traditional corporations.” — 00:18:46

“Akash is first and foremost a supercloud. A supercloud is a cloud that sits on top of private, public, hybrid — doesn’t matter — any cloud-capable computer, and gives a familiar user interface… whereas the compute comes from all over the place using a marketplace mechanism.” — 00:06:36

“Akash is world’s first supercloud that’s decentralized. A supercloud is essentially a cloud that sits on private, public, hybrid, single computer, and effectively provides you capabilities and features that would look like a normal cloud, and with a marketplace where you can source resources from places that you normally can’t find in your typical cloud marketplaces.” — 00:01:27

“We are in a very crazy political and social-economical environment. If we don’t decentralize, we will die as a society, and feudalism over freedom is going to prevail — and that’s the reality we are faced with right now.” — 00:08:38

“What that is trending towards is a feudalistic society that we work so hard as a humanity to escape — few people with the resource that everyone needs have control over those resources and charge a tax. All of us pay a cloud tax without even realizing we are paying a cloud tax: an average 50% of all the income that online services make go to a cloud provider, and these providers are unelected companies with shareholders, profits… as a primary driver.” — 00:07:11

“Not too many people know Akash is one of the largest IP address marketplaces… Akash is used by healthcare companies to store HIPAA compliant data. Akash is used by NSA — they’re doing a project inside the Department of Defense… Akash is so robust and so simple, it is composable, part of a larger protocol.” — 00:19:36

“Akash is the world’s first open supercloud. A supercloud is a cloud that sits on private, public — doesn’t matter — a cloud that essentially decouples the control plane from the resource plane… to take advantage of resources that are globally distributed.” — 00:02:17

“Protocols or systems that build upon these simple, permissionless, censorship-resistant primitives is key, because when you build on Akash the guarantee you get is: you own it, no one’s going to stop you, and you can change it.” — 00:20:19

“We believe in the unstoppable web, and what we’re doing with Akash is essentially creating a new operating system for the internet. And how we’re doing that is by tapping into unused 85% capacity in 8.2 million data centers… Akash is essentially a peer-to-peer network of computation clusters designed for global-first applications, planetary-scale applications.” — 00:02:59

“What AWS did for Amazon is what Akash is doing for the rest of the world.” — 00:02:52

“The goal really is to get to cloud parity, in the sense in a few years from now we should be able to go to Walmart or some large company and be like, hey, you can move all your stuff from Amazon to Akash. We’re not there yet — we need to build a lot of capabilities.” — 00:27:45

“There are lots of risks, but I think the biggest one would be — Akash is too early right now to be shut down by regulation. I think if regulation is out of the way, I think Akash is unstoppable. But if regulation comes and basically says you cannot exist, I think that’s going to be a big problem.” — 01:34:27

“We wrote a paper… the subsequent version, 2018, that introduced a notion of a supercloud — the Akash supercloud platform — where we said, well, you can decouple the control plane from the resource plane, where the resources could effectively be any cloud-capable compute.” — 00:08:21

“We quickly realized that you cannot scale to any level of high performance using a centralized control plane. The only way to really massively scale is go back to the roots, how internet was founded — a decentralized network.” — 00:05:08

“The goal really is to get to cloud parity, in the sense, in a few years from now you should be able to go to Walmart or some large company and be like, hey, you can move all your stuff from Amazon to Akash. We’re not there yet — we need to build a lot of capabilities.” — 00:25:48

“You cannot scale to any level of high performance using a centralized control plane. The only way to really massively scale is go back to the roots, how the internet was founded — decentralized… instead of a single control plane we can have a decentralized control plane where there’s no dependency on a single point of failure… blockchains happen to be the only data structure that would allow us to have a decentralized control mechanism, or decentralized coordination mechanism.” — 00:05:07

“We launched the mainnet in 2020 — first time ever a decentralized network could launch general purpose compute. There was nothing like it. The only thing that was close to Akash was Golem, and Golem was not general purpose… Akash was the first general purpose compute [where] you can actually demonstrate a Docker container running in a decentralized setting… they call it the Airbnb for data centers, without the cleaning fee.” — 00:09:19

“The goal really is to get to cloud parity, in the sense that in a few years from now you should be able to go to Walmart or some large company and be like, hey, you can move all your stuff from Amazon to Akash. We’re not there yet — we need to build a lot of capabilities.” — 00:25:46

“We want to be at least 500 providers by end of the year; end of next year we want to have at least a thousand providers… the goal is to get to 10 millisecond latency for 98% of active internet population… the kind of stuff you can do with such a low-latency network is beyond imagination.” — 00:45:43

“There was like 8.2 million data centers 10 years ago, which reduced to 7.5 [million] now, and increasingly concentration happening… but that’s reversing. We’re seeing smaller data centers now that are popping up everywhere with GPUs on them. So I don’t think hyperscalers are going to be the winner-take-all in this market.” — 00:37:48

“It’s already the largest distributed network in terms of regions — it’s got more regions than Amazons and Googles of the world. By end of the year we anticipate at least 500 to 600 new regions, so by end of the year it’ll be the largest distributed network in terms of physical distribution. So you can imagine the edge computer applications that can be finally enabled on Akash.” — 00:07:52

“The best analogy for Akash is Airbnb for data centers… you have about 80 providers now, all different sizes and shapes… and the way you know someone is telling the truth is through auditing. There is a decentralized network of auditors… [that] verify this compliance offline and post a proof online… Akash gives you more trust than traditional Amazon because Amazon will not show you the certificates.” — 00:07:55

“Akash is the world’s first supercloud. What a supercloud is, is essentially a cloud that connects any cloud-grade compute — be it in a data center, be it in a home, or be it in a private setting or a public setting — and brings [it] together in a way that’s easily consumable by developers around the world.” — 00:00:02

“Cloud infrastructure, which arguably is one of the most important layers that holds humanity together… It is an invisible layer that we don’t talk about much. I mean, all of us pay a cloud tax. 50% of all the monies we pay to an online service goes to a cloud.” — 00:13:25

“We tried open source cloud several times before, and none of them really took off or succeeded or even survived because they didn’t have the resources that you can actually acquire through open source. So for the first time, I had an opportunity to create not only open source software, but also give access to the resource in an open manner.” — 00:15:07

“A big portion of the roadmap current focus is to get to parity to Amazon web services. Post-parity is when, I think, Akash will start really seeing stuff that are not possible… How about, you know, 1,000 plus regions close to cell phone towers offering 10 millisecond latency…? What kind of possibilities do you unlock when you have such high-performance cloud? Well, you can do cloud gaming… that’s just the tip of the iceberg.” — 00:19:49

“What immediately comes to my mind when you say about Web3 is liberty. Right? The liberty that the Web originally promised but somehow got lost in the way and now we are bringing back the original values… Liberty, sovereignty, freedom.” — 00:00:51

“If Akash is successful, this is a multi-trillion dollar network you’re talking about — like an internet of compute… It’s a bold vision, but if they’re successful you guys are thinking too small.” — 00:53:46

“If we can decentralize at least the compute layer, we have a much better chance of decentralizing the entire stack. Versus if you don’t decentralize this base-level layer, we’re going to be having trouble.” — 00:47:18

“I’m always confident because Akash never breaks. It never does. It’s been around for 4 years. It is more alive than Amazon Web Services, and I take that.” — 00:15:48

“Really founded Akash to democratize cloud and compute, because compute is the fabric that connects the modern society, and that controlled by a few people that are unelected, that take tax from us, is not a future that I want for me.” — 00:02:53

“When you remove the need for a centralized scheduler or a centralized control plane… and you go to a fundamental peer-to-peer manner, you can scale unlimited… One of the reasons Akash has been around for 4 years and hasn’t had a single outage except for the planned upgrades is because it’s foundationally [decentralized]… During last one year Amazon went down 100 times — when it goes down you can’t order your Uber or your DoorDash. Is that the foundation that we want to build our AI [on]?… From a technological standpoint, I think there is no other way to scale at a global scale other than decentralization.” — 00:08:35

“The a16z did a study… that said something like 50% of your money that you pay for online services goes to a cloud provider that you have no idea who it is. Compute is the fabric that holds a modern society together, and compute is foundation for AI. If you think that you’re going to have to pay a tax to someone that you have no idea who it is, you’re not elected — we used to do that when society was feudalistic. So we need decentralization to avoid feudalism, that’s the simple answer. And we believe compute should be a common good; it should not be in the hands of few.” — 00:07:12

“We’re going to fix the supply chain problem at some point… but once the supply chain is fixed, you get the economy of scale that large hyperscalers have that may overpower the network. So I think we have maybe about 2 years to really decentralize, because we have one thing the hyperscalers don’t: Akash is the only place to get on-demand H100s right now — and that’s going to be not forever.” — 00:10:24

“It’s been online for about four years, has not gone down a single time, like not a single time. Amazon goes down 200 times a year, right?… because Akash is open source, because you have so many eyeballs watching this code base 24/7, the chances of bad code getting in thereby causing instability is very, very low.” — 00:19:50

“The white paper we wrote in 2017, 2018… introduced the notion of a supercloud. Supercloud, the term was coined by Cornell in 2015, but we implemented the supercloud. So Akash is the first implementation of an open source supercloud.” — 00:04:22

“As we keep delivering, the goal is to get to cloud parity. Cloud parity is when we can sufficiently say that you can actually move all your workloads from your Amazon of the world to Akash. We’re not there yet… our biggest capabilities, be it reservations or managed backends, I think will put us to the parity.” — 00:10:47

“Akash is introducing [a] Managed Services Marketplace. Now it transitions from a service that offers compute… to a network of networks… the creators of Postgres, or creators of MySQL… offering their services… and get a cut for their creation. So we’re empowering open source creators with a business model because open source cloud needs open source software to function.” — 00:29:26

“Decentralization — I think if you remove enough layers, you’ll see the biggest risk is not the actual software but actually people building the software and the roadmap. So who sets the roadmap and who delivers the roadmap, all that is done in the open. That’s a big part of Akash’s decentralization approach.” — 00:06:31

“It turns out there are three companies that control these computers, 80% of it, at least. Now, that, to me, is a little troublesome. Why? Because so much power in the hands of a few. You know, we had something like that. It was called feudalism, right? If you have so much power in the hands of a few, that few dictate the future of this public utility.” — 00:03:54

“Imagine a world where computation is free and accessible as the air we breathe. We believe compute should be a public utility — like air, like water, like all the resources that we don’t need to seek someone’s permission to use.” — 00:11:05

“Today if you use an online service, 50% of every penny you pay goes to one [of] the cloud providers… All of us, no matter what sector we are in, half of every dollar we spend goes to an invisible layer — or the cloud tax, we call it.” — 00:06:44

“Imagine a world where computation is free and accessible as the air we breathe. We believe compute should be a public utility — like air, like water, like all the resources that we don’t need to seek someone’s permission to use.” — 00:11:05

“Today if you use an online service, 50% of every penny you pay goes to one [of] the cloud providers… Look at companies like Asana, where 63% of the fees actually goes to one of the cloud providers; companies like Palantir… about close to 40% goes to a cloud provider. All of us, no matter what sector we are in, half of every dollar we spend goes to an invisible layer — or the cloud tax, we call it.” — 00:06:44

“Akash transitions from this simple cloud to this network of networks, which I think is very, very powerful, and I think that’s how we get to mass adoption.” — 00:41:50

“Cloud is ultimately a public utility that should not be closed off and they should not be this opaque, and that’s the reason why you want to decentralize a cloud.” — 00:11:32

“We all use a cloud and we all pay a cloud tax without realizing — somewhere north of 50% of every penny we pay to an online service goes to a cloud provider.” — 00:10:45

“Akash is the world’s first decentralized supercloud, and the core essence of a supercloud is where you can decouple the control plane from resource plane, where the resource plane can consist of any cloud-capable computer, orchestrated in a decentralized and open manner.” — 00:02:08

“We believe compute should be a public utility. Compute should be as free as the air we breathe and water we drink, and that’s the future we’re all building towards… the access should be ungated, [it] should be open, [it] should be public.” — 00:08:48

“The costs are getting out of control. Over 50% of every single penny we spend on online services goes to one of the three cloud providers. All of us pay a cloud tax without knowing who we [are] paying the tax to… even Department of Defense, which uses Palantir, clearly pays a cloud tax… Andreessen Horowitz did this study, it’s called the Trillion Dollar Paradox.” — 00:05:12

“Akash is the world’s first decentralized supercloud where it connects underutilized resources with people that need these resources… the obvious advantage is cost — anywhere from 80% to 50% cheaper than what you normally pay on Amazon — and more importantly access, like GPUs which are very very hard to get on your regular cloud. Akash offers advanced GPUs like H100s for like $150.” — 00:00:00

“Akash is the world’s first decentralized supercloud. Supercloud, like the name suggests, is a cloud that federates any cloud-capable compute bit, a private or public cluster… where users can effectively access compute using a marketplace mechanism at significantly lower cost, usually 80% cheaper than what you would normally otherwise pay on traditional cloud… one of the biggest users for Akash is Nvidia, and Akash is directly integrated into Nvidia products.” — 00:01:26

“Cloud, which is perhaps the most important fabric that holds a modern society together, and a very important public utility, should be as free as the air we breathe and the water we drink, and the access to which should not be gated. That was the whole premise of starting Akash.” — 00:38:17

“For us success we measure in market share, right? So our first milestone would be 1% market share of the cloud business.” — 00:39:01

“Akash has been around for about four years and has not gone down a single time, and you can’t say quite the same for networks like Amazon, where Amazon has reportedly about 200 outages on average per year.” — 00:10:59

“Now there’s enormous amounts of compute spread across hundreds and hundreds of data centers across the world… that really brings about the need for a layer to connect these distributed data centers, and that layer cannot be centralized. I think that’s critical for this layer to be decentralized, and Akash is one of the decentralized networks, if not the leading decentralized network, that’s connecting these distributed data centers.” — 00:13:05

“Picture this: in 1969 the first and most widely adopted decentralized network was created to withstand a nuclear attack, and today we end up with four companies controlling the internet… Have you really forgotten the very principles that gave birth to this digital age?” — 00:00:45

“Andreessen Horowitz did this study: half of every penny you spend on online services goes to a cloud provider — in some cases like Asana, where over 63% of all the fees you pay to the cloud service goes to a cloud provider. In a way, we’re paying this invisible tax, the society, all of us together — we don’t even realize we’re paying this invisible tax.” — 00:05:13

“I’m the co-founder for Akash Network. Akash is the world’s first decentralized supercloud, and it leverages underutilized compute to create a super cluster and offers compute at significantly lower cost in a permissionless and decentralized manner.” — 00:00:00

“In a world where we have heavy censorship, having a layer that is considered to be the public utility — right, compute — should be as free as the air we breathe and water we drink, and that is a foundation that humanity can accelerate on, and that foundation needs to be permissionless.” — 00:08:03

“The goal is to get parity with the current cloud… Akash will transition from a resource network to a services network where you have hundreds if not thousands of services all offered in a decentralized manner… The question I’ll leave you folks with is: what happens when you have thousands of regions that are spread across the world with thousands of services? Amazon has 32 regions right now — imagine that as a thousand regions.” — 00:19:39

“We all, all of us as a society, pay something like cloud tax without even realizing we’re paying it… 50% of every single penny you pay for online services goes to [a] cloud provider that is not elected, that’s not selected, that is there because the users really don’t have too much options.” — 00:32:27

“The whole point of building open source decentralized systems is to remove [single] centers of liability. Anything could happen [to] Overclock Labs, and Akash will run as designed — that’s the future we want. So we are very reluctant to collect user information.” — 00:18:06

“The only way to compete is to do something they cannot do. That’s why decentralization comes into play. They cannot go inside a home, where networks like Akash can go inside a home… You got to beat Amazon at a game that they cannot play — open source is something that Amazon can’t play.” — 00:13:46

“Akash has been alive for about four years; it has not gone down a single time. Whereas if you look at networks like Amazon Web Services, it goes down about 200 times a year… The reason why Akash is incredibly reliable is because there’s not a single company or a single cluster that controls the network, but rather a distributed network of nodes that interact with each other in a peer-to-peer manner without a central point of failure.” — 00:13:50

“Compute should be a public utility. Compute should be as free as the air we breathe and water we drink, so that innovators everywhere, from every corner of the globe, have the tools to build and create and transform without barrier… It’s very important that we return to sovereignty — control over what gets built, control over what goes and touches our daily lives.” — 00:10:14

“The big risk as a society as a whole we have is the return of feudalism. Feudalistic society is where few have power over the many… the few today are the ones that are running AI… There’s this discussion about artificial general intelligence: what happens when somebody cracks it and no one knows that they cracked it? Because there’s no incentive for them to share this AGI — they can very well go and play in the stock market or do things that are significantly advantageous to the few. So if we don’t do something about this layer, we go back to feudalism… I call the term digital feudalism.” — 00:08:48