Ben Goertzel, CEO of SingularityNET and head of the ASI Alliance, says the future of decentralized artificial intelligence will depend on more than better models or louder open-source rhetoric. If AI agents are to participate in large-scale autonomous commerce, blockchain systems must overcome deep structural limitations around decentralization, scalability, and security. In his view, decentralized AI can still challenge dominant technology companies, but only if its infrastructure evolves fast enough to support the next phase of machine-driven economic activity.
Agentic commerce is raising the bar for blockchain infrastructure
The debate is increasingly shaped by the idea of agentic commerce, a term used by Stripe co-founders Patrick and John Collison in their February 2026 annual letter. They described a future in which AI agents can independently discover opportunities, make decisions, and execute transactions without constant human intervention. Stripe’s assessment was optimistic about crypto’s long-term role in that transition, but it also made clear that current blockchain systems are not yet ready to serve as the main rails for this kind of autonomous economic activity.
According to the Stripe founders, the industry’s current limitations should be viewed less as fatal flaws and more as engineering constraints similar to those faced by the early internet in the 1990s. Their report identified cost predictability and transaction throughput as the most immediate obstacles. From Stripe’s perspective, blockchains may eventually need to support somewhere between 1 million and 1 billion transactions per second if they are to underpin a fully autonomous commercial environment.
Goertzel called that projection entirely plausible. He noted that even today, peak digital financial traffic already reaches into the millions of transactions when activity is still largely initiated by humans and routed through intermediaries. In a world of autonomous agents, the scale changes dramatically. A single user may no longer trigger a single financial action; instead, an entire team of AI agents could be operating continuously on that user’s behalf, each one generating its own stream of decisions and transactions.
Speed alone will not solve the problem
For Goertzel, the issue is not simply about increasing raw transaction speed. He argues that the blockchain systems required for agentic commerce must also address the broader architecture of coordination. That means balancing the classic blockchain trilemma of decentralization, scalability, and security while also making sure that intelligent agents are not trapped inside isolated single-network environments.
He pointed to several other capabilities that will matter in practice: handling the massive information flows produced by autonomous groups of agents, enabling direct peer-to-peer settlement, and supporting what he described as a more advanced form of decentralized identity. In other words, future blockchain infrastructure must become more than a ledger for payments. It must act as a coordination layer for complex machine-to-machine interaction.
That is why Goertzel does not believe the answer lies in a single monolithic network designed to do everything. Instead, he favors a system made up of specialized networks that can interoperate smoothly. He compared this model to a modern highway system, where dedicated lanes exist for buses, freight, and express traffic. By separating flows according to function, congestion can be reduced and overall efficiency improved. Applied to blockchain, that means a shard-like architecture in which different networks or network segments are optimized for specific tasks while remaining compatible with the broader ecosystem.
Decentralized startups versus the consolidation of AI power
Goertzel’s argument also sits within a larger concern about the direction of the AI industry. While blockchain emerged from a decentralized ethos, AI development is becoming increasingly concentrated in the hands of a small number of powerful corporations. These companies are investing billions of dollars into proprietary infrastructure, giving them outsized control over compute, data pipelines, model deployment, and distribution. The result, he suggests, is a growing risk that the future of AI will be shaped by corporate gatekeepers rather than open public processes.
Still, he does not see this concentration as inevitable. A broad ecosystem of smaller firms and decentralized projects is attempting to compete through a different strategy. Rather than trying to outspend the largest technology groups on their own terms, these actors are leaning on agility, niche specialization, and open collaboration. Their bet is that architectural diversity, transparent methods, and community-driven development can create forms of innovation that large centralized systems struggle to reproduce.
Goertzel acknowledged that decentralized AI organizations do not yet approach the size of firms such as Google or Microsoft. But he argued that they are reaching a scale at which meaningful competition becomes more realistic. What he described as one of decentralized AI’s “secret sauces” is diversity: the ability to draw together communities, researchers, AI algorithms, and datasets from around the world rather than forcing development through a single centralized pipeline.
In his telling, that diversity is becoming even more important as more leading researchers begin to question whether simply building larger large language models will naturally lead to artificial general intelligence. Goertzel said this skepticism is consistent with the direction his own organizations have pursued from the beginning, especially through the Hyperon approach to AGI and superintelligence.
Why open infrastructure and democratic governance matter
Goertzel also addressed the governance question directly. Asked how the risk of AI control by a few corporations or governments could be reduced, he argued that openness and decentralization must extend across the entire AI pipeline. That includes the large-scale deployment and operation of AI systems, fair data supply, the teaching of broad human values, and collective decision-making about how these systems evolve.
His position is that open-source code on its own is not enough. It has to be paired with decentralized infrastructure and decentralized governance if AI is to remain transparent and broadly accessible. Without that combination, open code could still end up running inside highly concentrated systems with limited accountability. With it, he believes AI has a better chance of delivering benefits widely rather than reinforcing existing power structures.
This framework also explains why SingularityNET and the AGI Society are organizing the AGI-26 conference. The event is intended to explore different interpretations and technical paths toward general intelligence, reflecting Goertzel’s view that no single model family or institutional structure should dominate the search for AGI.
The bigger takeaway for crypto and AI
The broader significance of Goertzel’s remarks is that they connect two major technology debates that are often discussed separately. One is whether blockchains can evolve from speculative financial rails into serious infrastructure for machine-native commerce. The other is whether AI can remain open and pluralistic as capital and control become more concentrated at the top of the industry.
In this framing, the future of decentralized AI depends not only on algorithmic progress but also on the maturation of the underlying transaction and coordination layers. If autonomous agents become meaningful economic actors, blockchains will need to support far greater throughput, more predictable execution costs, stronger identity frameworks, and seamless interaction across multiple networks. If they fail, centralized providers may become the default operators of agentic commerce.
Goertzel’s message is therefore both technical and political. Technically, he sees a path forward through interoperable, specialized blockchain architectures rather than all-in-one monoliths. Politically, he argues that decentralized AI must preserve openness, transparency, and democratic participation if it wants to offer a credible alternative to Big Tech. As the race toward AGI accelerates, that combination may determine whether the next generation of intelligent systems is controlled by a handful of dominant firms or shaped by a more distributed global ecosystem.