Top Ten Things to Watch in the Crypto AI World in 2025

Author: Teng Yan

Compiled by: Luffy, Foresight News

On a crisp morning in January 2026, you find a slightly shabby newspaper at your doorstep. Yes, it is printed on real paper. Somehow, it has survived the artificial intelligence revolution.

Flicking through the newspaper, you see a headline about AI agents coordinating global supply chains on the blockchain, while at the same time newly launched cryptographic AI protocols are vying for dominance. A half-page report introduced a digital "employee" hired as a project manager: this is so commonplace nowadays that no one would make a fuss.

A few months ago, I would have scoffed at the idea and might have even bet my portfolio that such progress would be at least five years away. But that’s how quickly crypto-AI will disrupt the world. I am convinced of this.

After recovering from a nasty stomach flu, I sat down at my desk to start the new year, wanting to start with something worthwhile: something that would spark curiosity and maybe even spark some debate. What could be better than trying to peek into the future?

I don’t usually rush into making predictions, but crypto AI is just too tempting to resist. There is no historical experience to follow, no trends to refer to, it is like a blank canvas to imagine what will happen next. To be honest, thinking about reviewing this article in 2026 and seeing how wrong I was makes it more interesting.

So, here’s my take on what crypto-AI might be in 2025.

1. The total market value of encrypted artificial intelligence will

reach 150 billion US dollars

Currently, crypto-AI tokens account for only 2.9% of altcoin market capitalization. But this won't last long.

As artificial intelligence covers everything from smart contract platforms to Memecoin, decentralized physical infrastructure networks (DePIN), and new primitives like agent platforms, data networks, and smart coordination layers, it is an inevitable trend for it to be on an equal footing with DeFi and Memecoin.

Why am I so sure?

• Crypto AI sits at the intersection of two of the most powerful technology trends I’ve ever seen.
• AI frenzy triggering event: An OpenAI IPO or similar event could trigger an AI frenzy around the world. At the same time, Web2 institutional capital is already paying attention to decentralized artificial intelligence infrastructure as an investment target.
• Retail investors are crazy: The concept of artificial intelligence is easy to understand and easy to get excited about, and now they can invest in it through tokens. Remember the Memecoin gold rush in 2024? The same craze is happening with artificial intelligence, except it's actually changing the world.

2. The resurgence of Bittensor

Nineteen.ai (Subnet 19) outperforms most Web2 providers in inference speed

Bittensor (TAO) has been around for many years and can be called the elder statesman in the field of cryptographic artificial intelligence. Despite the AI ​​craze, its token price has been languishing, unchanged from a year ago.

In fact, this "digital hive mind" has quietly made a leap: there are more and more subnets with lower registration fees, some subnets have surpassed Web2 similar products in actual indicators such as inference speed, and are closely related to the Ethereum Virtual Machine (EVM). ) compatible, introducing DeFi-like features to the Bittensor network.

So, why didn’t TAO take off? Steep inflationary plans and a shift in attention towards AI agents have hampered its development. However, dTAO (expected to be launched in the first quarter of 2025) may be a major turning point. With dTAO, each subnet will have its own tokens, and the relative prices of these tokens will determine how TAO is distributed.

Reasons why Bittensor is expected to be revived:

• Market-Based Token Release: dTAO ties block rewards directly to innovation and real measurable performance. The better the subnet, the more valuable its tokens will be, and therefore the more TAO releases it will receive.
• Focus on Fund Flow: Investors can finally direct their funds to specific subnets that they are optimistic about. If a particular subnet adopts an innovative approach to distributed training and achieves significant results, investors can show their support by investing funds into it.
• EVM integration: Compatibility with EVM will attract a wider community of native crypto developers to join Bittensor, closing the gap with other networks.

Personally, I'm looking at the various subnets and keeping an eye out for those that are making real progress in their respective fields. At some point, we will have Bittensor’s version of DeFi summer.

3. The computing market will become the next battlefield similar to L1

Jen-Hsun Huang: The demand for reasoning will “increase a billion times”

One obvious megatrend is the massive demand for computing.

Nvidia CEO Jensen Huang famously said that the demand for inference will "increase a billion times." This is an exponentially growing demand that will disrupt traditional infrastructure planning and urgently require "new solutions."

The decentralized computing layer provides raw computing power (for training and inference) in a verifiable and cost-effective manner. Startups like Spheron, Gensyn, Atoma, and Kuzco are quietly laying the groundwork to capitalize on this trend, focusing more on products than tokens (none of which currently have tokens). As decentralized training of AI models becomes feasible, the potential market size will rise dramatically.

Comparison with L1 blockchain:

• It’s like 2021: Remember when Solana, Terra/Luna, and Avalanche were competing to be the “best” L1 blockchain? We will see a similar free-for-all between computing protocols competing to attract developers to use their computing layers to build AI applications.
• Web2 Demand: The $680 billion to $2.5 trillion cloud computing market is bigger than the crypto-AI market. If these decentralized computing solutions can attract even a small percentage of traditional cloud customers, you could see the next wave of 10x or even 100x growth.

The risks are huge. Just like Solana emerged in the L1 blockchain space, the winner in the computing market will dominate a whole new frontier. There are three key elements to look out for: reliability, cost-effectiveness, and developer-friendly tools.

4. AI agents will flood blockchain transactions

Olas agent trading on Gnosis. Source: Dune/@pi_

Fast forward to the end of 2025, 90% of on-chain transactions will no longer be triggered by a human clicking a “send” button.

Instead, they will be executed by an army of AI agents that tirelessly rebalance liquidity pools, distribute rewards, or execute micropayments based on real-time data sources.

This doesn't sound far-fetched. Everything we’ve built over the past seven years—L1 blockchains, scaling solutions, DeFi, NFTs—is quietly paving the way for a world in which artificial intelligence dominates on-chain activity.

The irony? Many developers may not even realize that they are building the infrastructure for a machine-led future.

Why this shift?

• Avoid human errors: Smart contracts are executed strictly as coded. In turn, AI agents are able to process large amounts of data more quickly and accurately than human teams.
• Micropayments: These agent-driven transactions will become smaller, more frequent, and more efficient. Especially as transaction costs drop on Solana, Base, and other L1/L2 blockchains.
• Invisible infrastructure: Humans will happily give up direct control if it means less hassle. We trust Netflix to automatically renew our subscription service; then trusting the AI ​​agent to automatically rebalance our DeFi positions is a natural next step.

AI agents will generate amazing activity on the chain. No wonder all the L1/L2 blockchains are courting them.

The biggest challenge will be making these agent-driven systems accountable to humans. As the ratio of agent-initiated transactions to human-initiated transactions continues to increase, new governance mechanisms, analytics platforms, and audit tools will be needed.

5. Agent-to-Agent Interaction: The Rise of the Crowd

Source: FXN World Documentation

The concept of swarms of agents—tiny artificial intelligence entities working seamlessly together to execute grand plans—sounds like the plot of the next hit sci-fi/horror movie.

Today's artificial intelligence agents mostly operate in isolation, with minimal and unpredictable interactions with each other.

• Agent groups will change this status quo, enabling AI agent networks to exchange information, negotiate and collaboratively make decisions. Think of it as a decentralized collection of specialized models, each contributing unique expertise to larger, more complex tasks.
• One group might coordinate distributed computing resources on a platform like Bittensor, while another group could tackle disinformation and verify the source of information in real time before the content spreads on social media. Each agent in the group is an expert and performs its tasks accurately.

These swarm networks will produce intelligence far more powerful than any single isolated AI.

For groups to thrive, common communication standards are essential. Agents need to be able to discover, authenticate and collaborate, regardless of the underlying framework they are based on. Teams like Story Protocol, FXN, Zerebro and ai16z/ELIZA are laying the foundation for the emergence of the agent community.

This brings us to the key role of decentralization. Distributing tasks among groups based on transparent on-chain rules makes the system more resilient and adaptable. If one agent fails, the other agents immediately pick up the slack.

6. Crypto AI workforce will be a hybrid of humans and machines

Source: @whip_queen_

Story Protocol hired Luna, an artificial intelligence agent, as their social media intern, paying her $1,000 a day. Luna doesn't get along well with her human colleagues: she almost got one fired when she boasted about her superior performance.

As bizarre as it sounds, it's a sign of things to come. In the future, AI agents will become real work partners with autonomy, responsibility and even salary. Companies across a wide range of industries are exploring hybrid human-machine workforces.

We will work hand in hand with AI agents, not as slaves, but as equal partners:

• Productivity surges: Agents can process large amounts of data, communicate with each other, and make decisions around the clock.
• Build trust through smart contracts: The blockchain is an impartial and impartial overseer that never tires or forgets. The on-chain ledger ensures that important agent actions follow specific boundary conditions/rules.
• Social norms evolve: We will soon have to deal with the etiquette of interacting with agents. Should we say "please" and "thank you" to artificial intelligence? If they make mistakes, do we hold them morally responsible, or do we blame their developers?

I expect marketing teams to be the first to adopt this model because agencies are great at generating content and can live stream and post on social media 24/7. If you are building an AI protocol, why not deploy an agent on-premises to demonstrate your capabilities?

In 2025, the lines between “employees” and “software” begin to blur.

7. 99% of crypto AI agents will die, only practical agents will survive

We will see Darwinian selection among AI agents. Why? Because running an AI agent requires money in the form of computing power (i.e., reasoning costs). If an agent cannot generate enough value to pay the "rent", it will be eliminated.

Example of an agent survival game:

• Carbon Credit AI: Imagine an agent scouring a decentralized energy grid, identifying inefficiencies, and autonomously trading tokenized carbon credits. It can make enough money to cover its own computing costs. Such agents can thrive.
• Decentralized exchange arbitrage bots: Agents that exploit price differences between decentralized exchanges can continuously generate revenue to pay for their inferences.
• Spammers on X: Meanwhile, what about the virtual AI influencer who only tells cute jokes but has no sustainable source of income? Once the novelty wears off and the token price plummets, it will disappear and be unable to sustain operations.

The difference is clear: Pragmatically oriented agents thrive, while others fade away.

This natural selection is beneficial to the field. Developers are forced to innovate and prioritize productive use cases over flashy gimmicks. As these stronger, more effective agents emerge, they will silence the skeptics.

8. Synthetic data will surpass human data

It is often said that "data is the new oil". Artificial intelligence relies on data, but its huge demand for data has raised concerns about a coming data shortage.

Conventional wisdom holds that we should find ways to collect private real-world data from users and even pay them for it. But I’ve come to agree that in highly regulated industries or where real-world data is scarce, a more practical approach lies in synthetic data.

These are artificially generated datasets designed to simulate real-world data distributions, providing a scalable, ethical, and privacy-preserving alternative to human data.

Reasons why synthetic data is powerful:

• Infinite scale: Need a million medical X-rays or 3D scans of a factory? Synthetic generation can be produced in unlimited quantities without waiting for real patients or real factories.
• Protect privacy: No personal information is at risk when using synthetic datasets.
• Customizable: You can tailor your data distribution to your exact training needs, inserting edge cases that may be too rare in reality or difficult to collect due to ethical issues.

Granted, user-owned human data will still be important in many cases, but if synthetic data continues to improve in authenticity, it could overshadow user data in volume, speed of generation, and freedom from privacy constraints.

The next wave of decentralized AI may revolve around “small labs” that create synthetic data sets that are highly customized for specific use cases.

These small labs will cleverly circumvent policy and regulatory hurdles in the data generation process, just as Grass leveraged millions of distributed nodes to circumvent web scraping restrictions.

I will elaborate on this in a subsequent article.

9. Decentralized training will really make a difference

In 2024, pioneers like Prime Intellect and Nous Research are pushing the boundaries of decentralized training. We have trained a 15 billion parameter model in a low-bandwidth environment, demonstrating that large-scale training is feasible outside of traditional centralized approaches.

While these models are currently less practical (lower performance) compared to existing base models, I believe this will change in 2025.

This week, EXO Labs went one step further with SPARTA, cutting communication traffic between GPUs by more than 1,000 times. SPARTA enables large model training under low-bandwidth conditions without the need for dedicated infrastructure.

What struck me most was their statement: “SPARTA works on its own, but can also be combined with synchronization-based low-communication training algorithms such as DiLoCo for even better performance.”

This means these improvements can be stacked to further increase efficiency.

As technological advances like model distillation make smaller models practical and efficient, the future of AI lies not in size but in better performance and greater accessibility. Soon, we will have high-performance models that can run on edge devices and even mobile phones.

10. 10 new crypto-AI protocols will reach $1 billion in market

capitalization (not yet launched)

ai16z will soar to US$2 billion in 2024

Welcome to the real gold rush era. It's easy to think that the current leaders will continue to dominate, with many comparing Virtuals and ai16z to the early days of smartphones (iOS and Android).

But the market is too large and untapped for it to be dominated by just two companies. By the end of 2025, I predict that at least 10 new crypto-AI protocols that have not yet launched tokens will have a circulating (non-fully diluted) market capitalization of over $1 billion.

Decentralized artificial intelligence is still in its infancy, and there is a large pool of talent gathering.

We have every reason to expect the emergence of new protocols, new token models, and new open source frameworks. These new players can displace existing players through a combination of incentives (such as airdrops or staking), technological breakthroughs (such as low-latency inference or chain interoperability), and user experience improvements (no code). A shift in public perception can happen in an instant.

This is both the charm and challenge of this field. Market size is a double-edged sword: the pie is big, but for skilled teams, the barriers to entry are low. This sets the stage for a Cambrian explosion of projects, in which many will fade away but a few will become transformative forces.

Bittensor, Virtuals, and ai16z's dominance won't last long. The next batch of crypto-AI protocols with a market cap of $1 billion is coming. There are a lot of opportunities for savvy investors, which is why it's so exciting.

Bonus Highlight #1: AI agents are new applications

When Apple launched the App Store in 2008, its advertising slogan was "There is always an app for you."

Soon, you'll be able to say, "There's an agency for you."

Instead of clicking on an icon to open an app, you delegate the task to a dedicated AI agent. These agents are situationally aware, can cross-communicate with other agents and services, and can even initiate tasks on their own that you never explicitly asked for, like monitoring your budget or rearranging your travel schedule if your flights change.

Simply put, your smartphone home screen could become a network of “digital colleagues,” each with their own area of ​​work: health, finance, productivity, and social.

And because these are cryptocurrency-empowered agents, they can autonomously handle payments, authentication, or data storage using decentralized infrastructure.

Bonus Highlight #2: Robotics

While much of this article focuses on the software side, I’m also very excited about the physical manifestation of the AI ​​revolution: robotics. This decade, robotics will have its ChatGPT moment.

The field still faces significant obstacles, particularly in obtaining perception-based real-world data sets and improving physics capabilities. Some teams are rising to the challenge, using crypto tokens to incentivize data collection and innovation. These efforts are worth watching (think FrodoBots?).

I have been working hard in the technology field for more than ten years, and I can’t remember the last time I felt this kind of excitement from the bottom of my heart. This wave of innovation feels different: bigger, bolder, and just getting started.