Markus Levin: DePIN could power the AI agent economy

​DePIN remains a relatively little-known part of the crypto market, even though such projects could prove more useful than many popular tokens. They allow people to build networks of devices together and earn rewards for their contributions. In an interview with Traders Union, XYO co-founder Markus Levin discusses why the market often overlooks DePIN companies and how they could help shape the future of AI.

– You are the co-founder of XYO, one of the first DePIN companies. For readers who are new to the concept, how would you explain what DePIN is in simple terms? Where do you think the sector stands today in terms of its development? 

– DePIN is a way of building and operating real-world infrastructure through communities rather than through a single company. Instead of one organisation owning every sensor, wireless hotspot, or data collection device, thousands of independent participants contribute hardware and are rewarded with tokens for doing so. The result is infrastructure that is collectively owned, geographically distributed, and not dependent on any single entity.

We started XYO in 2018, long before DePIN had a name, so we have watched this sector grow from a niche idea into a category that serious investors and institutions are paying attention to. That said, it is earlier than the headlines suggest. The projects that will matter are the ones solving the hardest operational problem, coordinating thousands of independent participants at scale while maintaining data quality that businesses and institutions will actually rely on. That work is still ongoing across the sector, and the next few years will separate the projects with have built lasting infrastructure from the ones that only exist on paper.

– A number of financial scams have presented themselves as DePIN projects. For example, users were asked to pay a so-called sign-up fee, install software on their home computers, and “mine” tokens that had no real infrastructure behind them. What are the key signs that distinguish a legitimate DePIN project from a scheme like this? And what should the industry do to combat fraud? 

– The clearest sign of a legitimate DePIN project is verifiable infrastructure. If a network cannot show you the hardware, the data it generates, or an independent way to confirm that data is real, that is a significant red flag. Legitimate projects are also transparent about their token economics, do not require upfront fees to participate, and have a clear answer to the question of what the network actually does in the real world. As an industry, we need to hold ourselves to a higher standard around disclosure and third-party verification. The projects that are building real infrastructure should welcome scrutiny, because that’s ultimately what separates this sector from the schemes that have damaged its reputation.

– Despite XYO’s long history, its token is not among the largest cryptocurrencies by market capitalization. Why do you think DePIN projects with real-world infrastructure have attracted less attention than the more speculative segments of the crypto market? 

– Speculative tokens have always attracted attention faster than infrastructure does, because a token attached to a whitepaper can move on narrative alone, and for most of the last cycle that is what the market rewarded. Building physical infrastructure takes years, and the progress is harder to see from the outside, because it shows up as nodes being deployed, data being verified, and revenue coming in from people using the product rather than in a rising price chart. 

After the 2017 ICO boom, we watched many projects raise enormous amounts with no real business model or meaningful transparency, so we made a deliberate decision to build XYO as a real company with revenue behind it.

That’s why we are still here today. Because the projects that get through the down cycles are the ones with real revenue and a sustainable model rather than a story. The market has often rewarded valuation over fundamentals. We operate more than 10 million nodes and generated $8.8 million in revenue in 2024, more than the Filecoin network produced over the same period, despite its much larger market capitalization. Market capitalization reflects what investors are willing to pay for a token, but it doesn't necessarily measure the strength of the business.

We are not chasing short-term hype, and the company is built to generate revenue from real-world use rather than from a token narrative. Attention will come back to this part of the market, and when it does the projects that kept building through the slower periods will be the ones with something real to show for it.

– Is XYO a profitable business today? If so, where does most of the company’s revenue come from: products and services, or token-related activities?

– Yes. Our business is primarily supported by revenue from products and services rather than token-related activities or outside capital. XYO operates as a U.S. company with audited financial statements, and according to our SEC filings, we generated $8.8 million in revenue in 2024.From the beginning, our goal was to build a sustainable business rather than rely on token sales. That approach has allowed us to continue investing in the network while maintaining a business model based on commercial adoption.

– Let’s talk about AI agents. How fast does the underlying infrastructure need to be to support them effectively? For example, BNB Chain recently announced plans for a blockchain capable of processing up to 100,000 transactions per second. Is that a meaningful benchmark, or is it still not enough? 

– 100,000 TPS is an aspirational number to achieve, but high TPS alone is not the right benchmark for AI agent infrastructure. Agents are doing more than just moving value, they are observing, deciding, coordinating and verifying data continuously, and most of that activity never touches a blockchain directly. The issue for agents is not how many transactions a network can process in parallel. It is whether the underlying infrastructure can confirm a single action fast enough, reliably enough, for an autonomous system to act on it without waiting.

– Am I right in thinking that high transaction throughput alone is not enough for AI agents to operate efficiently? What other factors determine the performance of the infrastructure they rely on? 

– Yes, there are many factors and one of the main ones is interoperability. AI agents won't all operate within a single blockchain or ecosystem. They'll interact across different networks, cloud environments, enterprise systems, and real world devices. Infrastructure has to make those interactions seamless, otherwise agents spend more time waiting for information or resolving inconsistencies than completing useful work.

Security and verifiable data are equally important. An AI agent is only as reliable as the information it receives. If it can't determine whether a piece of data has been altered, spoofed, or comes from a trusted source, faster infrastructure doesn't solve the problem. As AI agents take on more responsibility, the networks that perform best will be those that can deliver trusted data and clear provenance alongside speed and scalability.

– How promising is autonomous AI agent-based trading today? Can AI agents already analyze markets, execute trades, manage risk, and interact with exchanges independently, or is this still largely an experimental use case? 

– AI agents can already analyze market data, monitor news and on-chain activity, execute trades through APIs, and adjust positions based on predefined risk parameters. We're also seeing major financial institutions experiment with agentic AI in trading and portfolio management, although human oversight remains a core part of those deployments.Where I think we're still early in fully autonomous trading is that markets are unpredictable, and agents need to distinguish between reliable and unreliable information, explain why they made a decision, and operate within clear risk controls. Those trust and verification challenges are just as important as the trading algorithms themselves. I expect AI agents to take on increasingly sophisticated roles over the next few years, but the strongest systems will combine autonomous execution with infrastructure that provides transparency, provenance, and accountability.

– How do you see the future of DePIN? What role do you think this sector could play within the broader crypto industry over the next few years?

– I think DePIN is moving toward becoming one of the foundational sectors of Web3 because it connects blockchains to real-world infrastructure. Over the next few years, we'll see decentralized networks supporting everything from location and connectivity to robotics, autonomous vehicles, and environmental sensing. As more industries rely on real-time data, the demand for infrastructure that is open, resilient, and globally distributed will only grow.More broadly, DePIN has the potential to change how infrastructure is financed and built. Instead of requiring billions of dollars upfront, networks can grow organically by rewarding people who contribute useful resources. That opens the door to faster innovation and broader participation. I also think DePIN will play an important role in the AI economy by providing trusted, verifiable data that autonomous systems need to interact with the physical world safely and reliably.