/biz/ - Business & Finance

>>20734310
Kaiko runs on Chainlink too anon.

>>20618479
They had a katacoda course for the V2 and V3, they didn't do one for the V4.
I hope they do some dev support hires, it's their main weakness right now.

>>20590720
That was the same argument given against moving outsourcing servers to cloud in the late 00s and early 10s.
Cost cutting is king, if worker pools are cheaper they will adopt, startups will do it first, then the big boys will follow.

>>20588570
I could see a startup focusing exclusively on helping businesses tokenize their datas with iexec and rent them on the marketplace or use them as collaterals for loans instead of shares.
You could even go further, imagine you decide to liquidate a company that failed to gain enough traction to make a buck, for example one that developed a fitness app or a travel app.
You tokenize the datas and put them on auction, let the market decide its price, there would probably a broker interested out there.

So many possibilities.

>>20587758
ERC721 is 2 years old already, it's the standard for non fungible tokens (Cryptokitties).
What iExec made is making data trading as simple as sending a token to another address.

>>20587141
Most collaterals today in DeFi are stablecoins (DAI, USDC, USDT).

Some stablecoins are collateralized themselves with other cryptocurrencies (DAI for example uses ETH and wrapped BTC among others).
Using datasets as collaterals is something totally new I've never seen even being talked about and I've still been trying to wrap my head around all the implications since they announced they were working on it back in March.

"Data is an asset" is not really new but up to this day there never was a way to really trade or rent it as a commodity, let alone tokenize it. This is huge and strange in so many ways.

>>20585478
I do not really care about partnership announcements desu, to me it's obvious all the big cloud companies will end up becoming worker pools since Alibaba and IBM joined.

Amazon already proposes services related to Hyperledger and Corda but it's almost only node hosting, everything that is HPC is simply not compatible with DLT without something like iexec. I do not think they will propose datasets anytime soon however, but compute capacity? I see them all in before end of 2021.

>>20584988
Dropbox support for results is nice but this is the dataset as NFT collaterals is the real innovation here, this is some uncharted territory stuff.

Well now it's in the open

https://medium.com/iex-ec/iexec-v5-interoperability-and-privacy-tools-for-defi-3cf674e8cd8c

>>20584399
The tech giant teased is Google, this was confirmed on the Telegram.
We yet have to see what they worked on, probably a paper like they did with Chainlink, we shall see.
Dropbox still is a unicorn and I am rather suprised they did something with iExec, this is their first involvement in crypto as far as I can say.

Someone made a mistake

I just did some crumb chasing and I found something, they released the V5 page of their SDK in clear, probably by mistake.

On the Github you have still the instruction for the V4
>https://github.com/iExecBlockchainComputing/iexec-sdk

But on the npm package page of their SDK they are at the V5
>https://www.npmjs.com/package/iexec
You can clearly see they have a "Storage" subsection that IS NOT on the github.

Guess what? Dropbox and IPFS are BOTH listed.
Dropbox confirmed.

>>20513481
It's just a side project for iexec, the point is that you get something supposedly ecological (I guess if you discard the manufacturing and wastes) since it's self sufficient.
They were partnered with Nerdalize too in the early days, it was a dutch startup using servers to heat buildings but they went under last year.

>>20512969
I have no idea if iExec will announce a partnership with Nvidia (that said they are both in the CCC already).
What I know is that Microsoft experimented with a TEE solution called Graviton (there is an AI chip with the same name made by Amazon btw) which is very similar to SGX by intel, simply it runs on GPUs.

CPUs are versatile and good enough to compute many types of tasks, however everything that is related to HPC and AI is vastly faster and more efficient when processed by GPUs.

At the same time I could see cloud providers join iExec marketplace for your run of the mill CPU centered tasks then keep the prenium stuff (FGPAs/AI chips) in their closed garden.

Otherwise iExec is yet to support AMD SEV (which is their SGX equivalent) but it's not really surprising since Lei Zhang worked for 5 years at Intel on confidential computing and Intel owns something like 95% of the hardware cloud market.

But wait there is more, in 2018 iexec sealed a deal with Wuhan (yes, the bat soup capital), they are going to integrate iexec in the smart-city infrastructure after the V5.

Today every city is full of cameras, traffic lights, meteo sensors with satellites monitoring them in almost real time. That’s where the IoT and smart cities memes enter the picture, with them you can have data redundancies that are independent and separated from the cars and can largely be used to improve real-time decision making.

The thing is the car manufacturers sell everywhere on the planets and all these city data points are controlled by tens of thousands of different entities, either private or public who come and go, these data are not necessarily processed either (it’s not because a camera films something that it’s doing real time object recognition for example). The yesterday solution is each manufacturer signing tens of thousands of contracts with these entities, setting up there own servers, etc… pretty much a logistical nightmare. Then you have the iExec solution, which is to transact and process these data on a global cloud computing marketplace in real time when needed, they’re so big brained that they even went as far as making a video giving us a glimpse:
>https://www.youtube.com/watch?v=VEh3JHivS10

Right now other manufacturers are finally understanding what’s happening, they began acquiring startups like Argo AI and Cruise. VW and Ford were in such a hurry they accepted to make an alliance on Argo AI and they’re throwing billions at it.

Car makers are like the phone makers from the last decade in the sense that their hardware and software are decoupling, the difference however is it’s not about user experience or features here, it’s ALL about the data and the first one to reach a certain level of safety will very likely end up being the hegemonic solution.

So maybe you’re thinking
>"they will let their cars run a lot and wait for the results then"
WRONG anon! The threshold for safety for self driving is much much higher than consumer faced software, you can get your social media account hacked or even your credit card number stolen, it’s not the end for you, if you crash into a truck on the exchanger because the sensor broke you’re probably dead or crippled for life. Data collected by the cars are not enough, it needs to anticipate by knowing what is out of sight.

Hello RLC bros, yesterday I wrote a bit about TCF and immediate use-case of iExec with Ethereum/Hyperledger/Avalon but today I wanted to write more about the big picture and the long term applications (>5 years) for a decentralised cloud computing marketplace.

Let’s talk about cars.
Currently if you follow a bit the automotive industry you know there are huge changes happening, probably the biggest since Ford revolutionized the industry with the Model T, and it is NOT the move from ICE to batteries.
I don’t want to go too deep in a controversial discussion about Tesla and its valuation/bubble, but the big money is NOT primarily investing in Musk boxes on 4 wheels, they are investing in the fact that Tesla has billions of miles of data to train their Autopilot model. Every time a Tesla drives 100 miles they upload few Mb of data to the motherbase, they’re doing that since 2014, most car makers only implemented data collection during the last 2 years or if before at much smaller scale than what Musk does.

Waymo (former Google Cars) has got all sort of valuations thrown at it during the last year in the same window as Tesla, all in the hundred of billions because even though they have a much smaller fleet than Tesla, they likely benefit from other data collection and processing initiatives (Google Map, Recaptcha, etc…).

Data is the new gold and processing power the new oil. Up to now the conundrum was that companies were accumulating a shitton of datas but they didn’t want a competitor getting their hands on them by proxy (think Uber selling its datas to Walmart who then resells in its back to Lyft or some much more complex fuckery).
A decentralised open and global marketplace with TCF solves everything, not only that but it solves a lot of regulatory issues too, like with GDPR because datas can be monetised without companies trading informations on individuals.


Now all of this could sound like pipedream… until you realise where exec positioned itself in the CCC and Avalon:
>What does iexec do?
>iExec: 4 contributors will develop Ethereum smart contracts, integrate TEE options that support most of the the mainstream programming languages and native applications, and improve Avalon easy-of-use for developers (https://wiki.hyperledger.org/pages/viewpage.action?pageId=16324764))
That's right, iExec controls the Ethereum gateway for all these applications to flourish.

But wait, there is more:

Today let's say you're a company like airBnB and you're using AWS to run your infrastructure, you're accumulating a lot of precious datas on where people go in vacation, what type of surface they rent, where they rent it, for how long, etc... Think a million of different parameters, In short a goldmine in terms of analytics for real estate companies.
There are two issues with this though:
1) AWS literally knows everything about it too since they have access to all the datas in clear when they process it
2) You can't really monetize the data-set without revealing it.

TCF solves this: now Bezos can’t look above your shoulders anymore and if he tries he will merely see gibberish, but in bonus you can monetize the data-set to everyone. A chinese tourism company wants to buy some real estate in Germany? They pay to use dataset to get a map of where they rent. McDonald is wondering which district has the highest concentrations of tourists whose profiles match the typical clients? They do the same.
I used big companies for you to get the idea, but the reality is that a shitton of startups and small to average companies would benefit from this for decision making.
The only practical environment is a decentralised global market, aka: iexec.

The thing is that you still need multiple workers to work on the same job and compare the resulting hashes, preferably you want to be sure these actors don’t collide so having one entity doing all the computation is a big no.
That’s where iexec enters the place because they provide a neutral ground where:
a) requesters (Dapp devs and users) can find multiple entities to process the same job simultaneously and aren’t relying on a single party
b) workers from all around the world can maximize their uptime by loaning their computing powers to whoever needs it
c) requesters benefit from the workers competing on a global marketplace to have the best price

>CCC (Intel, IBM, Microsoft, Google, Alibaba, iExec) and Hyperledger (Linux Foundation) converging
>Hundreds of companies in Hyperledger (Walmart, SAP, Baidu, etc...)
You know the drill about Hyperledger, but wait these are not simply IBM/Google/MS, it's IBM Cloud, Google Cloud and Microsoft Azure.

Now Hyperledger Avalon, what's this?
> Hyperledger Avalon is born from Trusted Compute Framework (TCF) which is a ledger independent implementation of the Trusted Compute Specifications published by the Enterprise Ethereum Alliance. TCF is solving an issue with privacy and data.
Avalon is the most sponsored and fastest growing brick of Hyperledger, it’s what all big companies look at while rubbing their hands.

What does it do?
Ethereum and Hyperledger are alright to compute few lines of codes. For example an Ethereum smart contract is limited to 24kb, which is absolutely nothing (that's 5 times less memory than the first Gameboy) and for a reason every computation done on Ethereum must be done on the 25 000 other nodes of the network all running on hardware with a different processing power.
The issue with that is that you can't put a ressource intensive Dapps on the blockchain, remember how something as small as Cryptokitty clogged Ethereum for example. So a decentralised youtube/spotify/netflix/etc… are NEVER going to happen with onchain computations.

What's the obvious solution to that? Having an external machine (called a "worker") that isn’t a node doing the heavy lifting (called a "job") offchain for the Dapp (the "requester"), but this brings another problem: how do you verify the computations can be trusted?
That's where TCF enters the picture because you can verify they executed the valid code and didn’t, mostly by verifying hashes, not only that but the datas remain encrypted through the processing at the hardware level, that’s what SGX is all about, the owner of the machine can’t peek at the datas if the owner of the datas doesn’t want to.

>>13253982
Definitely another advantage. It would become sort of a way to "buy" certain cryptos anonymously through buying RLC first.

>>13253363
>buying a glorified xbox360 and thinking its a good investment
HAHAHAHAHAHAHAHAHAHAHAHAHAHAH

>>13253890
I think waiting for new versions/news for iexec pumps is the wrong way of thinking about it.

I see iExec as one of those "well I didn't expect that" sort of thing. One day it'll probably pump extremely hard, garnering a ton of attention and mainstream media attention. It's pretty fucking relatively unknown at this point and it blows my mind.