Greg Osuri on Akash Network’s decentralized cloud computing

0:01i'm ashton addison from event chain for investment pitch media and the crypto coin show and today on blockchain interviews we have greg osuri the founder and ceo of akash network greg welcome to the show and thanks for taking the time to be here thank you ashton thanks super excited to be here yeah i'm really excited to dive into a cache network i know you have

0:21a lot of things going on in the industry right now so if we could just kick it off by starting with like a high level overview of what is a cache network that would be great so akash is essentially the world's first open source cloud that's self-sovereign non-custodian permissionless right and it it works by connecting a layer of interconnected data centers computer

0:45providers usually if you think about the landscape 85 or so of computing capacity in data centers premiums are new so akash tries to aggregate its computing power and provides a open source platform to its users where the developers get to enjoy cost savings anywhere from two to three x of what they would otherwise pay on the cloud today

1:11along with an incredibly modern developer experience a experience that's really designed for for needs of a developer in 2021 definitely that's great and are there a lot of problems that need to be solved right now in just the centralized cloud computing industry yeah so a big and really if you think about the big fours you know amazon google microsoft

1:37alibaba they're created for the previous generation of software and they you know these things have been around for 15 years and as they got bigger and bigger they forgot what the the real focus should be as the developer right and also uh the they've gotten so big and they built competitive uh modes around them it's extremely hard for new players to

2:03come and actually provide services in a way they create in horizontally and we all know what looks like if you look at the airline industry or if you look at the telecom industry it's not like people don't fly or people don't talk on the phone these industries just stopped innovating right because of all capitalization so oligopolies tend to uh create products that are

2:27misaligned sometimes to the customer base and they tend to create competitive modes that makes it prohibitive for new players to come and partake in in what should be a competitive marketplace right so that's essentially what we're looking at uh for the cloud and cloud if you think about it it is the fabric that runs the society i mean cloud is where all of our data is like

2:50it or not is stored and processed uh and it's increasingly becoming larger and larger considering the amount of data that the process is doubling every year so so and uh there seem to be no other really a solution to this problem um so that that core you know problem of the cloud getting disconnected and misaligned with the with the people and the uh the interests of the people

3:15it serves is what we are trying to fix with akash by creating an open source cloud because open source as we know is is uh is a key for the innovation that we have in software and and we're trying to extend that innovation uh so that we can have a more uh a liberal and a more uh aligned uh compute platform definitely and in terms of the targets that you're focusing on i

3:43know there's a lot of blockchain and d5 based applications that are being hosted you know on centralized cloud competing services and they're looking for decentralized solutions are you focusing on those industries specifically right now or are you really just able to host you know all kinds of applications around the whole world yes so it's a generic platform you can

4:03essentially run anything that you run on amazon and google on akash and we you know but since the applications are broad we like to stay more focused so that our teams can focus on the efforts and be more connected to the users we serve right so and the reason we chose to go with with the d5 focus and more blockchain focus is because this is a crowd that

4:26has ideological alignment with what we're uh what we offer as a product as well as this is a crowd that's crypto native so trying to train a non-crypto native cloud crowd on how to use tokens and how to secure it takes a lot of effort and one more on top of that uh d5 or decentralized applications when we started focusing on this field and a few years ago

4:50it was an ascent and had significant problems number one being you know censorship resistance right like a lot of a lot of what we do is not very welcome in the broader uh ecosystem if you look at centralized cloud providers right so um uh so the the reason we wanted to pick and choose our own people to help first was an obvious choice and and fortunately the space has been

5:16growing rapidly and also um in uh you know by staying closer to our users we can also offer our akash token itself and and actually use some of these products that our customers uh you know you you know offer so it's it's a significant alignment with d5 that said our bigger broader market is really the bigger cloud and the next best the most attractive uh

5:43area for us to focus is machine learning and computation every cloud because that segment is currently suffering with the immense cost problem if you look at and do some horrible throwing report recently on how uh and that stated essentially 20 of margins tend to go to the cloud right so you know it's it's extremely hard being an ei company the growth stage air company that's

6:10trying to survive in this increasingly prohibitive uh layer called cloud so uh that's attractive for us because akash one of the biggest benefits is cost savings right so the model that we envisioned for an ai company to function is more of a peer-to-peer cloud so to give you an example um since it's so expensive for you know to operate in the cloud a lot of these

6:35companies tend to bring in their workloads in in-house right so they have a server from the offices that has gpu clusters and they tend to tend to you know use these clusters but the problem is they only use these clusters maybe two or three hours a day while they're training these models and most of the time they're not using them and but when they want to use them

6:57they tend to peak their their usage and they tend to need more than they they have and that's when they still end up using the cloud right so it's not completely optimizing your cost by bringing in uh it's very inefficient because of usage pattern in a scenario like that they can install akash and offer this unused capacity and earn credits in exchange

7:18um which which is essentially a cash token and they can use these credits to for when they want to use a peak capacity so that's that's really when you you see providers becoming users and users becoming providers and that's very attractive for us and we had incredible demand uh from machine learning and that's uh that we like to see that uh be our next use case yeah that's great

7:41greg and that's a huge inefficiency and that's the case with a lot of things right now where you know you're only utilizing the full capacity for a small amount of time and otherwise you would just have these assets that are sitting there and to be able to capitalize on that and share resources as well is a huge advantage so that's great to hear that

7:59your customers like that 8.2 million data centers 85 not used so you can imagine the kind of kind of capacity out there yeah that's crazy and i was researching into a cache and i saw that you recently released the main net 2. now it sounds like a big milestone could you just give an update on you know what is the new upgrades to the mainnet2 mainnet2 essentially completes what we

8:24promised in a white paper so this is the world's first marketplace that we propose our white paper into 2017 that we released and we laid out the design uh for how a reverse auction marketplace where a cloud computing network would look like and this is a realization of our of that vision we had over two years of testing as you know this is a complicated

8:45product and uh two years of testing multiple test nets and you know days months of work that went into uh essentially launching our uh the promise of decentralized cloud the current uh offering uh it's a baseline offering in the sense that you can actually um ask for a workload and providers will bid on the workload and you know you the tenant the the lowest bid that matches all the

9:13requirements uh tends to win and you can deploy and it's a phenomenal system that does basic operations on deployment and uh the and also adds different little features that we didn't quite uh talk about in a white paper on like improving trustability as you know it decentralized networks especially something that's not verifiable platform like less

9:39like a smart contracts platform akash is not uh so we have the question of like whether can we get the resources that we have promised so so we added this this feature called audited attributes where where providers that are advertising what they have is audited by a third party third party and whatnot so that improves the trust in the decentralized networks where it is

10:02amazing and we you know we built um usability tools on top of that so it's it's not as obvious that you're actually interacting with a uh with a blockchain but rather feels like a more uh open source tool than a creating transactions in the ancient translation this sort of thing that's great and yeah i i i know um you sort of mentioned that it's this new uh ui is sort of like a

10:30uniswap experience now i thought that was really interesting comparison can you just sort of elaborate on what you meant by that so one of the least understood and least appreciative things in decentralized networks i think is user experience right i love unit swap even though you know i pay a premium both in gas fees or even like uh exchange fees i tend to use unit swap

10:55first before i go to centralized exchanges why think about the shared experience of swapping two tokens right so if you want if you have to go on a centralized exchange you go log in solve the capture you know hope you get the traffic lights right the number of traffic lights and then go and log in and then you you know you you swap you pay the fee and then when you well first to deposit

11:19this whole like gamut of things you go through to deposit things and then you swap and then withdraw with two verifications your email and then your your you know 2fa and then you log out the entire process is so cumbersome versus going to uni-swap and click click click you're done i mean you pay extra gas fee but sure you know i'm willing to pay the

11:40premium because of the comfort i think people tend to um not understand the value of comfort and that what unisoft showed us is like yeah you don't need to be fast or you know cheap you can actually provide an incredible user experience based on a decentralized network and that's a lot preferred than than normal and on top of that you know the the beauty of

12:04non-custodialism beauty of like i owning my assets and the beauty of um there's a security that i get from owning my own keys uh all that adds up to the experience of unison so imagine extending similar experience to the cloud akash today you don't you this there's no email addresses and there are no passwords there's no need for you to give your personal information that we all

12:28know the the kinds of things that happen these days um and the compromises we have to make to just to deploy something on the cloud no credit cards and all that stuff so it's beautiful right now the experience is command line but yes we will extend that experience over to the web but when when we do that it you will notice a uni-swap style deployment experience and that's coming

12:48very soon so that makes the network a lot more programmable a lot more composable and what we're really excited with ibc coming coming live is extending that composability to the rest of the ecosystem through profitability is extremely powerful so my had this vision of like owner-less web services like where people can actually own web services not not a single entity not a

13:14single human being but the world can actually own a web service that is managed by a dab and you know has a threshold signature for deploying and controlling all that stuff all that can be realized now with the decentralized cloud i'm really excited about yeah that is very exciting and i'm uh really interested to see how that plays out as you continue to grow

13:34the network and earlier you mentioned about you know capitalizing on those underutilized resources and rewarding those participants with a cache network token i just wanted to touch back on the token because i know uh there is a lot of intricacies within the platform that are using the akash token so if you just elaborate on a little bit on you know

13:53what are the main functionalities with the akash token and does it create a sustainable ecosystem within the akash network sure so the akash akt is a native token of akash network the first uh primary use for the akt token is to secure the network as you know akash is a sovereign independent network it is an l1 in its own right right it is a set secured by a validator set that

14:19essentially uh come to consensus in a proof-of-stake uh algorithm and uh the each token has one vote that's how you secure the network so it's a state of token number two it is a value of exchange now akash you have different properties called stream payments and you know escort escrows like payments and akash token provides those beautiful features at a cost model

14:46that's softened so we don't we don't really have to worry about the ethereum cell gas design or stuff like that and a third thing is it is used to create subsidies right so right now we bootstrapped the akash network and the security of the akash network using the inflation so without inflation there is no incentive for folks to stake the token without which we will not have networks

15:12so the subsidies can be created by having control over the monetary policy and what what where we are going is i wrote a paper called bootstrapping a free market by borrowing from the future essentially the idea that yes you have this inflation and you can use this inflation to subsidize some of the costs and that will in in turn can actually accelerate growth

15:36and this is not a new idea this is an idea that their teens proposed and how we you subsidize corn industry in the united states or how we subsidize textile industry in in india so when you tend to when you have an industry and you want to accelerate or you know growth in in that industry using inflation that means creating new money supply and using that

15:58that to actually subsidize or like taxation but usually it's you know it's the inflation rate that that we create is actually a good way to do so right so we kind of borrowed some of those ideas and implemented or proposed to to be implemented uh very soon uh so using that we can actually reduce the cost even further to 10x so wow it's of course an economic hack

16:21right so and the difference is actually coming coming from the inflation and you know it gets added to the thing but yes that's not like sustainable on a long term but for whatever um in the sense the there's a finite amount of tokens you can't just keep printing tokens right and again the inflation is on a decay curve so the inflation goes down every block

16:43period um right now we we began as 100 million tokens and there will be 389 million ever created in eight years or so right so so it's not it's not uh infinite i mean it's still finite but as long as the ex the the subsidies are accelerating growth we which in turn increases the utility of the token which in turn attracts liquidity right so it goes in a

17:08cycle of like this growth feeding into the adoption and fading into the liquidity so and that's the beauty of this receptacle economics of akash great thanks for that greg and we're running short on time but one of my last questions is in terms of the long-term success of akash network what do you think will be one of the major key factors to achieving that long-term success

17:33the key factors will be product market fit the self-uh you know self-served user right a user that can actually use the network without talking to anybody without asking anybody in sustain that that is going to be the pinnacle of uh russia now we saw a certain amount of product market fit in our test nets i'm not going to claim that we have product marketplace because we just

17:58launched yesterday and and that is going to be our number one priority and again with product market fit you have to make adjustments and you have to make iterations on the way and we're really looking forward to we're entering this we're launching this network with the mindset that yes we're going to have a lot of things that we're going to learn and we're really excited about these

18:20learnings and actually working with the people to to get to the product market definitely it is very exciting and for the viewers that are looking to follow along with the updates to akash network and get involved what's the best way for them to learn more the best way of course go to our website our car network and you'll find a lot of resources our blog is very very

18:40active and consistently published but twitter is where we live so akash. akasha's twitter account is at akashnet underscore and my twitter account is greg osuri and we both are very very active on twitter and you know i respond to people really quickly on twitter and my dm's are open as well and of course their telegram channels and discord channels as well that are pretty

19:08heavy uh in terms of engagement great thank you greg i will leave those links in the description box below for the viewers all the best with akash network moving forward and let's follow up in the near future thank you ashton this was a really really fun