Author of the Opinion: Gaurav Sharma, CEO of io.net
Artificial intelligence may still be in its early stages, but it has already brought significant scientific and technological breakthroughs in developed countries. Unfortunately, these developments come at a cost: the dangerous centralization of AI.
In Forbes' list of the top 50 private AI companies for 2025, all companies are located in developed countries, with 80% in the United States.
AI still favors capital-rich tech giants in developed countries.
For many emerging economies, the threshold to enter the AI revolution is out of reach. We need to ensure that innovation and AI development are accessible to the broadest range of projects.
At the core of the issue is access to computing resources. Training and deploying large AI models require massive GPU computing power. Supply cannot keep up with demand, driving the price of Nvidia H100 chips to over $30,000.
An ambitious AI research company may spend 80% or more of its funding on computing—resources that could have been used for R&D or talent. Well-funded tech giants may raise billions to acquire these resources. Other regions of the world cannot do the same.
The consequences are profound. AI-driven innovation risks becoming a monopolistic technology controlled by a few companies and countries. Promising AI applications in agriculture, education, or healthcare in developing economies may never materialize—not due to a lack of talent, but because of limited access to computing resources.
From a geopolitical perspective, the insufficient supply of computing resources is beginning to resemble oil or silicon. Countries without sovereign access to computing resources will be forced to import, creating dependencies on countries that may not align with their national goals, and exposing them to foreign energy and real estate markets. These dependencies threaten economic competitiveness and national security.
If access to computing resources remains concentrated in developed countries, influence will be as well.
Cutting-edge AI technologies, from large language models to diffusion models, will be shaped by the same perspectives, narrowing diversity and embedding systemic risks. Developing countries face the risk of being locked out of contributing to or benefiting from the technologies that define the global economy.
Centralization ensures that disproportionate returns flow to those with privileged access, leaving smaller players behind, often those building locally relevant tools. Over time, barriers to competition in the AI market may become unstable oligopolies, freezing developing countries out of critical industry transformations. Centralized control of infrastructure always produces distortions, and AI will be no exception.
The solutions to the challenges of accessibility and centralization are surprisingly simple: blockchain-driven computing markets. Just as Uber unlocked idle cars and Airbnb unlocked spare rooms, decentralized computing markets unlock underutilized hardware. The result is lower prices and a more diverse, resilient ecosystem of suppliers and consumers.
Globally, millions of GPUs are idle in data centers, enterprises, universities, and homes. By aggregating these GPUs into on-demand clusters through blockchain, underutilized hardware is offered at a fraction of the cost of centralized computing. Startups in low-income countries can afford to scale AI workloads, no longer shut out by the capital advantages of industry leaders.
Without blockchain, this model would be impossible. Tokens serve as a coordination and trust layer, incentivizing reliability in decentralized physical infrastructure networks (DePIN). Leading DePINs require computing suppliers to stake tokens to incentivize reliability and penalize downtime. Developers pay with tokens, enabling seamless cross-border settlements.
For hardware providers, tokenized rewards create a fairer economic model: compensating computing owners based on usage, providing previously unattainable income without sacrificing core purposes. For developers, the opportunity to access cheaper computing incentivizes participation and innovation in AI. This creates a positive feedback loop— as more participants join the decentralized computing market, computing becomes more affordable and abundant.
Some critics argue that decentralized computing does not perform as well as hyperscale cloud providers, citing latency and quality concerns. The reality is quite the opposite. DePINs offer competitive performance in latency, concurrency, and throughput. Technologies such as intelligent workload routing, mesh networking, and high-availability tokenized incentives help maintain performance and dynamically optimize based on workload demands.
Moreover, certain DePINs have already built transparent network browsers that allow developers and investors to verify performance claims in real-time. These mechanisms help make DePINs more reliable and cost-effective than traditional providers.
DePINs also offer a more diverse range of products than hyperscale cloud providers. Currently, over 13 million devices are online, allowing developers to leverage a wide spectrum of hardware to find the right tools for their AI projects, from high-performance cloud-grade GPUs to specialized edge devices.
We have a narrow window to define the technological landscape for generations to come. While many companies in the U.S. and China may already be ahead, decentralized computing markets offer a promising alternative. By lowering costs and expanding access, startups, scale-ups, researchers, and enterprises around the world can compete more equitably. Emerging economies can build models for their own languages, healthcare systems, cultural beliefs, and financial needs.
The question is not whether decentralization is necessary, but how to enable global developers to access this opportunity while increasing the number of companies listing their surplus computing on DePINs. Only through decentralized computing can AI truly become accessible and serve as many people as possible, rather than just entrenched oligopolies.
Author of the Opinion: Gaurav Sharma, CEO of io.net
Related Reading: U.S. Energy Secretary Proposes Accelerating Grid Access for AI and Bitcoin Mining Companies
This article is for general informational purposes only and is not intended to be legal or investment advice, nor should it be construed as such. The views, thoughts, and opinions expressed here are solely those of the author and do not necessarily reflect or represent the views and opinions of Cointelegraph.
Original Article: Opinion: Decentralized Computing Networks Will Democratize Global AI Access
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
