Article by: David Marsanic | Originally posted on DailyCoin
At first glance, the large but unassuming building appears as nothing but a factory or giant warehouse. Spanning 255,000 square feet, the location is not a fulfillment center or a warehouse for shipping online orders but a key strategic point in Amazon’s bid for domination.
The Ashburn Data Center is packed end-to-end with high-end computer hardware that powers much of the internet today. This giant facility is just one of at least 40 Amazon data centers in Northern Virginia, fulfilling the computing needs of many US businesses on the East Coast.
In the West, a further eight were in San Francisco and Seattle each, and seven in Oregon. In Ireland, the big tech cloud service providers had to stop their expansion plans as the country’s electric grid couldn’t handle their demands for electricity.
These huge facilities power the “cloud,” a business that accounts for billions in revenue. What is more, this crucial infrastructure puts Big Tech giants Amazon, Microsoft, and Google in the position to control much of the Internet.
Why Businesses Need the Cloud
According to one industry survey, roughly 90% of all organizations use the cloud. What is more, they report that the cloud handles 75% of all their computational workload.
Cloud services are essential for many businesses with an online presence. Every company that wants to run a website requires computing power. Traditionally, this was done by running their own servers, even on personal computers.
However, as the website reaches more traffic, server requirements become higher. Buying hardware to scale is incredibly expensive and requires highly specialized staff to manage it. For that reason, cloud computing is a popular option. Rather than buying their own infrastructure, companies can rent from the cloud and pay as they go.
Thanks to its compelling business proposition, cloud computing became one of the most profitable businesses for Big Tech. In the first quarter of 2024, Amazon’s cloud service brought in $25.04 billion in revenue, or 17% of the company’s total revenue. Its impact on profitability was even higher, as AWS accounted for about 62% of Amazon’s total operating income.
In the same period, both Google and Microsoft saw accelerated growth in their cloud business. Critically, AI has been one of the main drivers of demand. However, the centralization of power in the cloud is already leading to undesirable outcomes, which brought about the rise of alternative, decentralized solutions.
Problems with Centralized Cloud
The concentration of power among the big tech, especially Amazon’s AWS, is not a result of an accident. Instead, the company is accused of predatory pricing, keeping prices artificially low to squash competition. Thanks in part to this practice, the big tech giants are steadily moving towards converting this control to financial gain, often at the expense of their customers and, ultimately, all Internet users.
Already, many major companies are seeing major downsides in the pay-as-you-go model. Once the companies scale, and rely on a large number of requests, the cloud can become prohibitively expensive. That’s why more and more companies are looking for alternatives, even buying their own infrastructure. Netflix did so in 2016, and Apple in 2017.
They found, however, that migration was not easy. For instance, Netflix had to pay AWS $15 million to manage its migration, while Apple paid a whopping $50 million for the same service. Pinterest and Airbnb also paid millions for the same reason.
What is even worse, centralized cloud computing also comes with significant security risks. Any type of server misconfiguration can lead to data leakage. Moreover, data providers and law enforcement agencies legally have free reign over all the data on the cloud.
This is something Web3 companies and competing cloud providers are hoping to change. Both see a significant advantage in decentralization as a challenge to the traditional model.
How a Decentralized Cloud Works
Decentralized cloud is Web3’s take on computer hardware. Instead of being in the hands of a few giant corporations, hardware is controlled by many providers. This ensures that no single provider has control over the network.
Thanks to blockchain technology, these networks can be self-managed. Instead of relying on centralized authorities, users and providers would interact with the network in a permissionless, pre-programmed way, ensuring a level playing field for all users and providers.
This would also have significant technical advantages, proponents of decentralization say. I would have a significant positive impact on network speed, enable high-end games on cheaper devices, and even tap into the immense computing power of billions of personal devices worldwide.
Although it may seem like a lofty idea, many companies are taking it seriously. The most immediate advantage of a decentralized cloud is its startup costs, as the necessary hardware is incredibly expensive.
One of these is Aethir, a decentralized cloud provider using third-party hardware to run its cloud. “If you want to deploy… in another region, you need to purchase the hardware… transport that hardware into that region,” says Mark Rydon, the CEO of Aethir. “Then, you need to have the correct entities and legal structures,” he adds, as well as a dedicated team.
This has brought multiple cloud startups to work on the decentralized cloud, integrating permissionless blockchain tech and taking advantage of the comparative benefits of distributed infrastructure.
Decentralized Cloud is Faster
Network speed is one of the main focus areas for current decentralized cloud providers, as they aim to leverage their unique advantages. In the decentralized cloud, infrastructure is more distributed.
Hardware can be much closer to the users, as Harrison Hines, CEO & Co-Founder of Fleek, a decentralized edge computing platform, points out. This means faster loading speeds and less online lag.
“If you were to build a cloud platform from scratch today,” he claims, you would start with a system that delivers data based on the users’ location. Performant movement of data, he explains, is key for all web services.
This is usually called the Edge, a system where giant servers are replaced by a network, enabling each node to be much closer to the users. Despite Big Tech’s current advantages, this makes a compelling case for competition.
“Yes, Google has all the data,” Hines explains. “If your website loads in three seconds instead of one, you lose a third of your customers. If it loads in five seconds, instead of one, you lose 90% of your customers.”
This speed and latency advantage will only improve in the future, says Mark Rydon, the CEO of Aethir, a decentralized cloud computing provider.
“As the network gets bigger, it gets faster,” Rydon explains, because there is an “increased likelihood that a user is… close to one of these nodes.”
Decentralized Cloud Enables Powerful Games on Cheap Devices
There are many potential use cases for the decentralized cloud. One of them includes gaming, where the high-quality connection makes high-end games playable on lower-end devices. The concept is not new, as Google unveiled its own cloud gaming platform, Stadia, in 2019. However, due to a lack of commitment from the parent company, Stadia shut down just two years later.
Other companies are not so quick to ditch the idea. One of these is Aphone, a decentralized cloud computing platform for mobile, especially focused on gaming. This “low-cost operating system … that can be compatible with every device” is very interesting for developing regions, says Aphone CBO William Paul Peckham.
“We are very focused on … developing markets,” including “Latin America, South-East Asia, places where people do not have the hardware to support … heavy games,” Peckham said.
The decentralized model allows the company to be run at a low cost. Aphone partners with Aethir and charges only $20 per month, which Peckham claims is enough to keep it profitable.
The potential is even bigger the more decentralized the network becomes. This may soon be the case, as decentralized cloud companies see the need for public blockchains to form the backbone of their network. They even see the potential for a network where each phone or laptop is its own node, in a decentralized network.
Decentralized Internet Powered by Phones and Laptops
The vision of a decentralized internet, where personal devices like phones and laptops contribute to a global, decentralized network powered by blockchain technology, is becoming increasingly feasible.
The potential computing power available from personal devices is immense, and much of it goes unused. So much so that there are initiatives where users can donate their unused computing to science. For example, IBM allows users to install an app that runs in the background when their device is idle and solves complex equations, helping to cure cancer.
Aethir CEO Mark Rydon says there is great potential in that route, even with some unique challenges, as users can turn off their devices whenever they want.
“Maybe we aren’t able to get full efficiency from these machines,” Rydon concedes. “I know categorically that we will find a way to kind of unite that latent compute power that does exist on consumer GPUs in a meaningful way.”
Companies like Aethir also realize the potential of decentralization on a network level powered by public blockchains. “We very much see ourselves as on the road to increasing decentralization,” Aethir CEO Mark Rydon revealed DailyCoin.
“We anticipate a fully permissionless compute ecosystem where any hardware can permissionless upload and connect to the network,” he added. Other projects are already working on leveraging personal computing, including apps that let users “mine” crypto while letting their devices idle.
However, significant challenges remain. These include technical problems and business interests interfering, both outside and within the decentralized cloud space.
Decentralized Cloud Is Open, But Not ‘Free
Despite its advantages, the decentralized cloud still depends on resources that have a real cost. Computing power is not free, whether in distributed servers or personal devices. Chris Hood, former Head of Digital Business Strategy and Platforms at Google Cloud, related some of these costs to DailyCoin.
Even with “mining” apps that convert idle devices into servers, users must pay the cost of electricity and hardware if they use their own devices.
“There’s a fee and a cost somewhere,” Hood points out, regardless of whether a network runs on the user’s own device or not. Because of that cost, there is a real challenge in getting users to buy into the network.
This also spills over to the incentives for owning hardware. If low-powered devices could perform as well as high-end ones, with the power of the cloud, there would be very little incentive to buy high-end devices. This also means that the devices out there would not have much computing power left over to give to the network.
Is Decentralized Cloud Really Decentralization?
Another major issue that needs to be addressed is decentralization on a network level. Few crypto projects are really decentralized, claims Chris Hood. And with Big Tech unwilling to open up its clouds, it’s not clear whether there is a viable path towards decentralization.
He points to the apps that offer incentives for users to keep their devices running, passively “mining” crypto. The problem with these devices is that someone else generates all the profit.
“So really what you’re asking me to do is become a miner where I’m not going to reap all of the rewards for my effort,” he explains. Companies running these services have their own costs and resources to maintain.
“If you’re telling me here’s a crypto app that I can set up, and 100% of anything I mine in the process is mine, awesome, I’m interested,” he explained. “But that’s not what we have.”
As Aethir CEO Mark Rydon points out, getting buyers is even more challenging. These tend to be enterprise customers who are reluctant to trust personal devices, which are at the whims of their users.
“It’s really, really, really difficult, if not semi-impossible today, to put those GPUs into a package that is sellable to an enterprise compute buyer,” Rydon says. Without enterprise buyers, the economics of the mining apps would break down. With no one paying into the system, the mined tokens will inevitably decrease.
Could AI Assistants Make Decentralized Cloud Work?
Still, there are potential solutions for many of these issues. Instead of selling to enterprise customers, Hood points out that apps could go the “Napster route,” outsourcing their computing power directly to their users.
“The reason why so many people participated in Napster and its peer-to-peer network was because they had a payoff,” the ability to share music online. Users wouldn’t mind sharing computing power and might not even expect monetary rewards if the service they get is valuable.
Hood points out one potential type of app that could go the “Napster” route: AI assistants. These emerging apps could soon be essential for everyday life, but most currently use the Big Tech cloud. Notably, OpenAI’s ChatGPT is powered by Microsoft’s Azure platform, costing the company billions of dollars.
Hood says that instead of being on the cloud and feeding all user data to its models, the more “natural” place for AI assistants is on users’ own devices. From there, they could “talk” to other assistants through a decentralized network.
One Path to Decentralized Cloud
The example of AI assistants is interesting as it gives a real use case for participating in decentralized networks. In time, more apps may start leveraging the power of their user’s devices. This includes social networks like X content platforms like Netflix and more.
All of these platforms have either a huge cloud bill or expensive hardware to maintain while also having a loyal user base that might be happy to share their computing with other users.
As many companies would benefit from this, there is an advantage to making this compute-sharing open and permissionless and brought together on a blockchain network. From there, smaller apps, websites, or individuals could tap into its network, breaking the monopoly of the cloud providers.
In that world, Google or Amazon’s data centers would be just one of the many providers competing for the same network on equal terms, with every person having a phone and some time away from its screen.