Over the past few months, due to the rapid growth of the AI industry, a large number of crypto industry professionals have shifted toward AI. Researchers with experience in both fields are also exploring a proposition that has never been successfully realized:
Can blockchain become part of AI infrastructure?
Over the past two years, the integration of AI and crypto has seen numerous iterations—AI agents, on-chain reasoning, data markets, and compute leasing. While hype has been high, very few projects have truly established a commercial闭环. The reason is simple: most remain at the “AI application layer.” But Gensyn enters the most critical and expensive layer of the AI industry:
Model training
How is this achieved? By organizing globally dispersed GPU resources into an open AI training network, where developers submit training tasks and nodes provide computing power; the network verifies training results and distributes incentives. What truly matters behind this is not "decentralization" itself, but rather an increasingly undeniable issue in the AI industry:
Computing power resources have rapidly consolidated into the hands of a few giants; major companies have already secured GPU supplies years in advance. Over the past year, the AI industry has increasingly established a clear trend: whoever controls GPUs controls the pace of AI development, especially in the era of large models, where training resources have become a core barrier to entry.
The supply of H100s remains tight, causing cloud service prices to continue rising. For major domestic companies, the first step in advancing AI is not expanding teams, but securing computing resources—this is why OpenAI, Anthropic, and xAI are all tied to major cloud providers, as the competition among models has essentially become a competition over infrastructure. The significance of Gensyn lies in:
A new way to organize resources for AI training
It targets the most fundamental infrastructure layer of the AI industry.
Many AI + Crypto projects focus on application-layer narratives—in short, everyone is just building apps. But Gensyn directly enters the training phase, which is the most technically demanding and resource-intensive part of the entire AI value chain, and also the layer most likely to create platform moats. Once a training network reaches scale, it becomes not just a compute marketplace, but potentially a critical gateway for future AI development. This is why the market continues to pay close attention to Gensyn—and why a16z has twice stepped in with major lead investments.
Second, it offers a more open model of computing power collaboration.
Traditional AI training heavily relies on centralized cloud platforms, which offer stability but come with continually rising costs—especially for small and medium-sized AI teams, where training resources have increasingly become a key constraint on innovation. Gensyn’s approach is to bring more idle GPUs into the network, enabling dynamic allocation of training resources and thereby improving overall computational utilization. This concept mirrors the logic behind the early emergence of cloud computing: not reinventing computation, but reorganizing computational resources. If this model can be sustained, it promises not only cost optimization but also a potential boost in resource efficiency across the entire AI industry.
Three, the technical barrier is precisely its important moat.
The real challenge in training networks has never been “connecting GPUs,” but rather: how to verify training results, how to ensure nodes honestly execute tasks, and how to maintain training reliability in a distributed environment. Gensyn has long been focused on solving exactly these issues—including probabilistic verification mechanisms, task distribution models, and node coordination systems. These elements may not be as flashy as the Agent narrative, but they determine whether the network is truly usable. In many ways, Gensyn is more like a deep tech infrastructure company—and this is what sets it apart from most other projects in the space.
Four: A commercial closed loop has already been established
One of the biggest controversies in the crypto industry has been that many projects have compelling narratives but lack real demand. However, AI training is different—it’s a proven, rapidly growing market with continuously expanding global demand for AI training, and a persistent shortage of GPU resources. Gensyn is entering precisely this well-defined segment of an existing supply chain. In other words, it’s not building “blockchain for blockchain’s sake,” but rather addressing the AI industry’s inherent need for a more flexible and open resource orchestration system. This is why an increasing number of investors are turning their attention to AI infrastructure: compared to short-cycle applications, infrastructure, once it achieves network effects, typically has a much longer lifecycle.
Finally, an interesting shift is taking place. In the past, people generally thought: Crypto is a financial system, and AI is a technological system.
But now, the boundary between the two is becoming increasingly blurred: AI requires resource coordination, incentive mechanisms, and global collaboration—areas where crypto excels most, enabling training capabilities to no longer be confined to a few giants, but instead evolve into a more open and collaborative system. At least for now, this is no longer just a conceptual story—it is evolving into a true AI infrastructure. And historically, the most valuable companies of the AI era have often emerged from the infrastructure layer.