Why has the infrastructure for Crypto Assets failed to keep pace with its ideals?

Author: Yohan Yun Source: cointelegraph Translation: Shan Oppa, Golden Finance

The periodic service interruptions and capacity pressures of centralized cloud infrastructure have created opportunities for companies to build distributed networks.

Proponents of distributed solutions argue that spreading workloads across multiple small nodes can reduce centralized risk. They suggest that this model could be highly valuable in industries with high computing demands and zero tolerance for downtime, such as artificial intelligence, gaming, and finance.

Carlos Lei, co-founder and CEO of Uplink, a connection market based on Decentralized Physical Infrastructure Networks (DePIN), told Cointelegraph that over time, as the performance of decentralized infrastructure reaches or exceeds that of centralized clouds, reliance on a single service provider will naturally decrease.

In today's technological landscape, decentralized infrastructure typically refers to blockchain - designed to achieve trust distribution and reduce single points of failure through decentralized verification and data storage.

However, the underlying infrastructure that supports access to these networks still largely relies on centralized cloud platforms.

Dependence on the Three Giants and the Risks of Centralized Cloud

According to data from Synergy Research Group, in 2024, Amazon Web Services (AWS), Microsoft Azure, and Google Cloud (collectively known as the “Big Three”) accounted for about 68% of global cloud infrastructure revenue, becoming the default hosting environment for many enterprises and blockchain applications.

(With the rise of generative artificial intelligence services, cloud service revenue has significantly increased Source: Synergy Research Group)

But this centralization means that any service provider experiencing a power outage, price adjustment, or regional capacity limit could trigger a cross-industry chain reaction.

Nökkvi Dan Ellidason, CEO of the technology infrastructure company Gaimin, told Cointelegraph that during the Amazon Web Services (AWS) outage in October, platforms like Snapchat, Roblox, Fortnite, and Kindle were all down, and financial services institution Coinbase was also severely affected.

These large platforms reduce upfront costs by providing subsidy programs for startups. For most companies, moving away from centralized cloud means taking on all the building and maintenance work that large service providers have handled for years, and the costs will be higher.

(Startups Credit Program provides thousands of dollars in cloud service subsidies Source: Amazon)

Despite these trade-offs, some companies have begun testing alternatives through “modular transformation.” Filecoin and Arweave are being used to store data and metadata outside of traditional clouds, while Akash Network and Render Network provide decentralized computing and GPU rendering services.

Ellidason stated that after the outages of Amazon Web Services (AWS) in October and Microsoft Azure in November, large enterprises are migrating some services to more resilient networks. They may still retain services from Amazon Web Services (AWS), but will first migrate storage operations, and may later migrate AI-related services.

The reliance of blockchain on centralized clouds undermines the essence of decentralization

The design purpose of blockchain networks is to decentralize trust and eliminate single points of failure, but a significant portion of the underlying infrastructure still operates on the same centralized cloud platforms. In major proof-of-stake networks, a large proportion of validators run on commercial cloud infrastructure rather than independent hardware.

A study released at the American Computer Security Association security seminar in August found that among traceable Ethereum validators, Amazon Web Services (AWS) is the largest single hosting provider, accounting for about one-fifth. Additionally, an analysis by Messari in 2023 showed that most validators and staking hosting services use centralized servers.

(Despite announcing a ban on cryptocurrency-related services the previous year, Hetzner remains one of the main hosting service providers for blockchain nodes in 2023. Source: Messari)

Lei stated that many operators choose centralized cloud platforms such as Amazon Web Services (AWS) or Google Cloud because they offer predictable performance, convenient setup processes, and reliable uptime.

These platforms have become the backbone of many Web3 networks as they simplify the complexity of infrastructure scaling and maintenance.

This dependency is often invisible to users, but it becomes apparent when AWS experiences outages. During the AWS downtime on October 20, Coinbase users reported issues with logging in, trading, and withdrawals, while Robinhood traders faced execution delays and API errors.

The integration of hybrid cloud model and blockchain has taken shape

The pressure on centralized cloud systems is increasing, driving a new round of infrastructure projects to explore distributed alternatives. These networks aim to leverage consumer-grade hardware, regional data centers, and idle computing power to gain capacity.

For example, Gaimin obtains GPU computing power from gaming computers and supplements it with small regional data centers. Ellidason stated that this approach helps to build a geographically distributed network, avoiding a single region from becoming a point of failure.

Uplink adopts a similar concept and applies it to the bandwidth sector. Its market allows individuals and local operators to sell surplus bandwidth to applications that require more coverage, reducing reliance on large telecommunications operators.

Some researchers and node operators believe that the future will be a hybrid model rather than fully decentralized — integrating massive cloud, edge networks, and bare metal servers, where traffic will be automatically rerouted when a certain area fails.

Magma Devs co-founder and Lava Network contributor Yair Cleper stated that the cloud will not disappear, and its resilience is crucial. Teams can start with a simple model and diversify as business needs grow, without having to rebuild the entire system.

As the demand for computing power continues to outstrip cloud capacity, this shift is not about abandoning platforms like AWS, but rather about enhancing the fault tolerance of systems and creating space for small regional infrastructure service providers to fill the gaps left by large data centers.