Written by: Conflux

On May 31, 2026, the U.S. Department of Commerce released new export control guidelines: the channel for Chinese companies to procure NVIDIA advanced chips through overseas subsidiaries in Malaysia and other places is officially closed.

In the same month, the President of Kenya halted the construction of a $1 billion geothermal data center involving Microsoft—because once completed, it would consume one-third of the country's electricity. President Ruto's exact words were: "This is equivalent to shutting down half the country."

Meanwhile, Huawei announced last week that the Ascend 950PR chip has entered mass production, with annual AI chip revenue expected to reach $12 billion.

Three events, three continents, three entirely different news stories. But they point to the same emerging reality: the competition for computing power is no longer just a matter for the tech industry.

A new oligopoly era is forming

In the past two years, there has been an often overlooked reality in the AI industry: although there appears to be a blooming variety, the underlying resources are increasingly concentrated.

The current AI industry chain can roughly be divided into four layers: GPU chips, cloud computing platforms, foundational models, and application ecosystems. In each layer, control is consolidating among a few players: in the GPU field, NVIDIA has almost become the only choice; in cloud computing, AWS, Microsoft Azure, and Google Cloud dominate; in the model layer, OpenAI and Anthropic have captured the vast majority of the high-end model market.

In other words: the same group of companies is simultaneously controlling chips, cloud platforms, models, and distribution channels. Eric Posner, a law professor at the University of Chicago, describes this phenomenon as the "AI Octopus," where the tentacles of these companies cover the entire AI industry chain.

This differs from the platform monopolies of the internet era—where internet platforms control traffic, AI platforms control intelligence itself. This kind of "oligopolistic monopoly" brings profound systemic risks:

• Concentration of control and pricing hegemony: A few companies hold sway over the pricing of AI, API access, and content moderation standards. Developers and enterprises face serious risks of "platform lock-in," where giants can change the rules or cut off access at any time.
• Infrastructure vulnerability: Highly centralized computing power is prone to single points of failure, easily causing widespread outages (such as large-scale cloud service failures), and places unbearable pressure on the electric grid and energy resources of any single region.
• Geopolitics and power hegemony: Computing power is shifting from a neutral infrastructure to a strategic bargaining chip. Due to export control restrictions, countries lacking independent computing capabilities (especially in the Global South) risk being marginalized and further widening the technological gap in this wave of technology.

In the future, more and more businesses will rely on AI to complete development, operations, customer service, marketing, and even decision-making. Once intelligence becomes a production tool, the importance of its control will far exceed that of search engines and social media.

Deepening "AI Iron Curtain"

In the past two years, the U.S. operations regarding chip export controls have become increasingly fragmented. During the Biden administration, an "AI diffusion rule" was established, dividing global cooperation into three tiers; after Trump took office, this rule was revoked in favor of case-by-case approvals and temporary bans. In response to this iron curtain, the reactions from various countries have been vastly different.

Saudi Arabia has directly declared 2026 as the "Year of Artificial Intelligence": through its sovereign fund's HUMAIN company, Saudi Arabia invested $3 billion in Musk's xAI, one of the conditions being to establish an AI data center of over 500 megawatts in Saudi Arabia; the UAE is constructing a 5-gigawatt AI park in Abu Dhabi—marketed as the largest globally outside the U.S.—with the first phase launching this year; in May, the UAE received the first batch of the latest NVIDIA chips exported from the U.S.

The logic of the Gulf countries is quite straightforward: the previous era relied on selling oil, while this era relies on buying computing power.

The EU’s anxiety comes from another direction: official data shows that over 80% of digital services in Europe run on non-EU infrastructure. The ongoing "Cloud Computing and AI Development Act" (CADA) aims to triple Europe’s computing power by 2030. France’s Mistral released a strategic document this April titled "European AI: A Playbook to Own It."

However, the most challenging circumstances exist for those economies that may not even qualify to compete: Kenya's $1 billion data center has been halted; Malaysia has allocated about $490 million to build a sovereign AI cloud. India is subsidizing GPU usage fees for researchers; Indonesia is preparing to develop local large models—these investments are not insignificant within their respective economic sizes.

But this year alone, the combined AI capital expenditure of Microsoft, Google, Amazon, and Meta is around $750 billion. This magnitude of disparity is part of the problem itself.

Moreover, the end of the competition for computing power keeps pointing to a more fundamental variable: electricity. The energy consumption of an AI inference task can reach up to 1,000 times that of traditional web searches. In anticipation of global data center energy consumption reaching 1,050 terawatt-hours by 2026, tech companies have even begun directly purchasing nuclear power plants.

Is there a possibility of being "non-aligned"?

It is against this backdrop that decentralized AI (DeAI) has started to receive attention. It attempts to answer the question: besides leaving the future to a few tech giants or a handful of countries, is there a third possibility?

If the internet can connect the global network through open protocols, can AI also connect global computing power through an open network? Can the idle GPUs, independent developers, research institutions, and enterprise data centers worldwide form an open AI infrastructure network?

The core idea of DeAI is not complicated: coordinate independent participants through open protocols to achieve an AI system without a single power center’s control. Combined with blockchain technology, cryptoeconomic incentives, and cryptographic validation mechanisms, it addresses the trust issues in anonymous networks and directly responds to the pain points of centralized AI:

• Breaking market concentration: Establish a network of distributed computing power, data, and model providers to form a market pricing mechanism based on free competition.
• Alleviating physical limitations: Distribute massive energy demands across the power grids of the world.
• Freeing from geopolitical dependencies: Build an infrastructure layer transcending any single jurisdiction, enabling the possibility of "sovereign AI."
• Increasing verification transparency: Use provable technical means to replace blind trust in the reputations of tech giants.

Supporters believe this model can reduce dependence on a single supplier, improve system resilience, and provide participation opportunities for small and medium countries and enterprises.

At the same time, the attitude of institutional investors is shifting from curiosity to substantive investment. Venture capital firms (such as DCG, a16z, etc.) are injecting hundreds of millions of dollars into DeAI protocols; traditional companies (e.g., Deutsche Telekom) are beginning to participate in the network as validators; additionally, some governments (such as Kazakhstan) are exploring connecting their idle national supercomputing resources to decentralized computing markets.

Conclusion

As the "State of DeAI 2026" report states, the core value proposition of DeAI is not about outperforming centralized systems in performance today, but about providing a foundational architecture that resists monopolies, rejects censorship, and decentralizes power.

With the decreasing costs of dedicated AI hardware (ASICs) and the continued prosperity of open-source models, the time window for DeAI to solve operational challenges has begun to open. The work of building the foundation for DeAI has just begun.

Of course, DeAI still has a long way to go before becoming mainstream. Whether in terms of performance, stability, or business models, it is still in the early stages. However, its significant meaning may not lie in immediately challenging OpenAI, but in providing an alternative solution.

Historical experience tells us: when an industry has only one choice, the question is often not whether power will be abused, but when it will be abused.

And the existence of competition itself acts as a form of check and balance.

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。