Opinion: Gaurav Sharma, io.net CEO
Artificial intelligence may still be in its infancy, but it is already driving significant scientific and technological advances across the developed world. Unfortunately, these developments have come at the cost of a dangerous centralization of AI.
Forbes’ 2025 Top 50 Private AI Companies list is all based in developed countries, with 80% in the United States.
AI remains skewed toward well-capitalized tech giants in developed countries.
For many emerging economies, the price of entering the AI revolution is unaffordable. We need to make innovation and AI development accessible to a wide range of projects.
AI access imbalance
At the heart of the problem is access to computing. Training and deploying large-scale AI models requires tremendous GPU power. Supply has struggled to keep up, pushing the price of Nvidia’s H100 chips to more than $30,000.
Ambitious AI research companies can spend more than 80% of their funding on computing. Those resources should be devoted to research and development and human resources. Deeply funded tech giants may raise billions of dollars to secure them. The rest of the world can’t do that.
The impact is far-reaching. AI-driven innovation risks becoming a proprietary technology controlled by a few companies or nations. Promising AI applications in agriculture, education, and healthcare in developing countries may never materialize, not because of a lack of talent, but because of limited access to computing.
Geopolitically, computing shortages are beginning to be reflected in oil and silicon. Countries without sovereign access to computing will be forced to import computing, creating dependencies on countries that may not align with national goals and exposing importers to foreign energy and real estate markets. These dependencies threaten economic competitiveness and national security.
The dangers of centralizing AI influence
If access to computing remains concentrated in developed countries, it will also have an impact on developed countries.
From LLMs to pervasive models, frontier AI technologies are shaped by the same perspectives, narrowing diversity and embedding systemic risk. Developing countries risk being excluded from contributing to and benefiting from the technologies that define the global economy.
Centralization flows disproportionate benefits to those with privileged access, leaving smaller players, often those who build locally relevant tools, behind. Over time, the barriers to competition in the AI market could become a destabilizing oligopoly, holding developing countries away from critical industrial transformation. Centralized infrastructure control has always created distortions, and AI is no exception.
Balance scale with distributed computing
The solution to the accessibility and centralization challenge is surprisingly simple. It’s a blockchain-powered computing marketplace. Just as Uber unlocked idle vehicles and Airbnb unlocked spare rooms, decentralized computing marketplaces unlock underutilized hardware. The result is lower prices and a more diverse and resilient ecosystem of suppliers and consumers.
Related: Bitcoin network will be used to secure election results for local Republican convention
Millions of GPUs sit idle in data centers, enterprises, universities, and homes around the world. By pooling these GPUs into on-demand clusters via blockchain, you can leverage underutilized hardware at a fraction of the cost of centralized computing. Startups in low-income countries can now afford to scale their AI workloads and are no longer locked out by the capital advantages of industry leaders.
The important role of blockchain
Without blockchain, this model would not be possible. The token is the coordination and trust layer that aligns incentives across the Decentralized Physical Infrastructure Network (DePIN). Leading DePINs require computing suppliers to stake tokens for reliability, with penalties for downtime. Developers pay with tokens, enabling seamless cross-border payments.
For hardware providers, tokenized rewards create fairer economics. This means compensating compute owners based on their usage, providing previously unavailable revenue without sacrificing their core purpose. For developers, access to cheaper computing will encourage participation and innovation in AI. This creates a beneficial feedback loop. As more participants join the distributed computing market, computing becomes more affordable and abundant.
Tackling challenges
Some critics argue that distributed computing does not perform as well as hyperscalers, citing latency and quality concerns. The reality is completely different. DePIN provides competitive performance across latency, concurrency, and throughput. Technologies such as smart workload routing, mesh networking, and tokenized incentives for high availability help maintain performance and dynamically optimize based on workload needs.
Additionally, certain DePINs are building transparent network explorers, allowing developers and investors to verify performance claims in real-time. These mechanisms help make DePIN even more reliable and cost-effective than traditional providers.
DePIN is more diverse than hyperscaler products. With over 13 million devices online today, developers can find the right tools for their AI projects with a wide range of hardware, from high-performance cloud-grade GPUs to specialized edge devices.
Leveling the playing field for AI
There is a narrow window in which we can define the technological landscape for future generations. While many US and Chinese companies may be in the lead, decentralized computing marketplaces offer a promising alternative. By lowering costs and expanding access, we create a more level playing field for startups, scale-ups, researchers, and businesses around the world. Emerging countries can build models that suit their language, health care system, cultural beliefs, and financial needs.
The question is not whether we need decentralization, but how to involve developers from all over the world in this opportunity, while at the same time increasing the number of companies that list their excess computing on DePIN. Only by decentralizing computing will AI be truly accessible and able to serve as many people as possible and not just entrenched oligopolies.
Opinion: Gaurav Sharma, io.net CEO.
This article is for general informational purposes only and is not intended to be, and should not be taken as, legal or investment advice. The views, ideas, and opinions expressed herein are those of the author alone and do not necessarily reflect or represent the views and opinions of Cointelegraph.