Author: Teng Yan, manager of Chain of Thought; Translation: Golden Finance xiaozou

2026 1 One clear, clear morning in September, you find a newspaper on your doorstep—yes, a newspaper printed on real paper that has somehow survived the AI ​​revolution.

Scroll through and you'll find a headline about AI agents coordinating global supply chains on the blockchain, and the newly launched Crypto AI protocol is vying for it Dominance. A half-page article describes the story of a digital “worker” hired as a project manager—something so commonplace these days that it barely attracts attention.

A few months ago, I might have laughed at this idea, and even thought that such development and progress would take at least 5 years to achieve. But Crypto AI is about to overturn the world at an alarming rate. I have no doubt of this.

As I sit down at my desk to start the new year, I want to start with something worthwhile—something that sparks curiosity and maybe even Stuff that sparks controversy. What could be more priceless than a peek into the future?

I don't usually get into predictions, but Crypto AI is just irresistible. There’s no past experience to look to, no trends to fall back on—just a blank canvas for people to imagine what’s coming next. Honestly, it would be interesting to think about revisiting this issue in 2026 and seeing how outrageous I was.

So, in this article, I will talk about my views on 2025...

1. The total market value of Crypto AI will reach US$150 billion< p style="text-align:center">

Crypto AI token is currently only 2.9% of the altcoin market cap.

With artificial intelligence covering everything from smart contract platforms to memes, DePIN and new primitives such as agency platforms, data networks and intelligent coordination layers, it will surely be able to embrace To have the same rise as DeFi and meme coins.

Why am I so confident about this?

Crypto AI is the fusion of two of the most powerful technology trends I have ever seen.< /p>

AI Frenzy: The OpenAI IPO or something similar could trigger a global AI frenzy. Meanwhile, web2 institutional capital is already surrounding decentralized AI. Infrastructure seeks investment targets

Retail Mania: The concept of artificial intelligence is easy for people to understand and get excited about, and now people can invest in this field through tokens. Remember the meme coin gold rush in 2024? It will be the same mania, The difference is that artificial intelligence is truly changing the world

2. Bittensor renaissance

p>

Bittensor (TAO) has been around for many years. It is a true OG. But its token price has been languishing, hovering at the same level as a year ago, despite the enthusiasm for artificial intelligence.

Under the surface, this "digital hive mind" has quietly made a leap forward: more subnets, lower registration fees, subnets that are better than Web2 in real indicators such as inference speed, EVM compatibility Introducing DeFi-like functionality to Bittensor’s network

So why isn’t TAO soaring? Dramatic inflation plans and a shift in attention towards agent platforms have held it back. However, dTAO (expected to be released in Q1 2025) could be a big turning point. With that dTAO, each subnet will have its own tokens, and the relative prices of these tokens will determine how token emissions are distributed

Why Bittensor is destined for revival:< /p>

Market-based emissions: dTAO ties block rewards directly to innovation and actual measurable performance. The better the subnet, the more valuable its tokens are—and therefore, the more emissions it earns .

Focused capital flows: Investors can finally target specific subnets they trust if a specific subnet is decentralized using an innovative distributed training method. Ying Erout, where investors can deploy capital to represent their views.

EVM integration: EVM compatibility will form a broader crypto-native developer community within Bittensor and build a bridge with other networks.

Personally, I've been paying attention to various subnets, especially those that are making real progress in their fields. Sometime in the future we will definitely usher in the Bittensor version of DeFi Summer. As I write this, TAO is priced at $480.

3. The computing market will usher in another wave of "L1 wars"

Hindsight Ming: An obvious big trend is the insatiable demand for computing power.

NVIDIA CEO Jensen Huang once famously said that the demand for inference will increase "a billion times." This exponential growth disrupts traditional infrastructure plans and calls out “we need new solutions.”

The decentralized computing layer provides raw computation (for training and inference) in a verifiable and affordable manner. Startups like Spheron, Gensyn, Atoma, and Kuzco are quietly building strong foundations, seizing the opportunity to focus on products rather than tokens (none of these companies currently have tokens). As decentralized training of AI models becomes feasible, the market size will rise dramatically.

Compare L1:

Like 2021: Remember Solana, Terra/Luna and Avalanche Is there fierce competition to be the “best” L1? We will see a similar free-for-all between computing protocols, as each protocol fiercely competes for developers and AI applications to use its computing layer.

Web2 demand: The cloud computing market size of US$680 billion to US$2.5 trillion far exceeds the Crypto AI market. If these decentralized computing solutions capture even a small share of traditional cloud clients, you will see the next wave of 10x or even 100x growth.

The stakes are huge. Just like Solan won in the L1 warJust like victory, this time the winner will dominate a new field. It is recommended that you focus on these three areas: reliability (for example, a reliable service level agreement or SLA), cost-effectiveness, and development-friendly tools.

4. AI agents will flood blockchain transactions

By the end of 2025, 90% On-chain transactions will not be triggered by a human clicking "send".

On-chain transactions will be executed by a group of AI agents who constantly rebalance liquidity pools, distribute rewards, or execute micropayments based on real-time data feedback.

This may sound far-fetched, but it is not. Everything we have created in the past seven years - L1, rollup, DeFi, NFT - has quietly paved the way for the on-chain world of artificial intelligence.

Ironically, many builders may not realize that they are creating the infrastructure for a machine-dominated future.

Why did this change occur?

No more human error: smart contracts execute exactly as coded. AI agents can process large amounts of data faster and more accurately than human teams.

Micropayments: These agent-driven transactions will become smaller, more frequent, and more efficient. Especially as the transaction costs of Solana, Base and other L1/L2 are trending downward.

Invisible infrastructure: Humans will gladly give up direct control if it reduces trouble. We trust Netflix to automatically renew our subscription; trusting the AI ​​agent to automatically rebalance our DeFi positions would be a natural next step.

AI agents generate a staggering amount of on-chain activity. No wonder all L1/L2 are catering to them.

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

5. Agent-to-Agent interaction: the rise of agent groups

The concept of agent swarms—tiny AI entities that seamlessly coordinate to execute a grand plan—sounds like the next big movie The plot of a popular sci-fi/horror movie.

Today’s AI agents are mostly lone wolves, operating in isolation with few unpredictable interactions.

Today’s AI agents are mostly lone wolves, operating in isolation. p>

Agent swarms will change this, enabling the exchange of information, negotiation and collaborative decision-making between networks of AI agents. Think of it as a decentralized collection of specialized models, each serving a larger, more Contribute unique expertise to complex tasks

The possibilities are staggering. An agent swarm can coordinate distributed computing resources on a platform like Bittensor. Another agent swarm can resolve misinformation and verify the source of content in real time before it spreads on social media. The agent swarm Each agent in the network is an expert in performing tasks accurately.

These agent swarm networks will produce much greater intelligence than any single AI. /p>

For an agent community to thrive, common communication standards are crucial. Agents need to be able to discover, authenticate, and collaborate, regardless of their underlying framework. The likes of Story Protocol, FXN, Zerebro, and ai16z/ELIZA The team is laying the foundation for the emergence of agent groups

This highlights the key role of decentralization. Distributing tasks among a group of agents managed by transparent on-chain rules will make the system more resilient.

6. The Crypto AI working team will be A hybrid of humans and AI

Story Protocol hired Luna (an AI agent) as their social media intern, paying her $1,000 a day. Luna didn’t get along well with her human colleagues—she almost fired one of them when she boasted about her superior performance. .

This may sound strange, but it is a harbinger of a future where AI agents become true collaborators, with their own autonomy, responsibilities, and even salaries. Companies across industries are working on human and AI Testing with mixed teams of agents

We will work hand in hand with artificial intelligence, they are not our slaves, but our equals:

Productivity Surge: Agents can process large amounts of data, communicate with each other, and make decisions around the clock without sleep or coffee.

Trust from smart contracts: The blockchain is a supervisor who is not partial, does not complain, and never forgets things. It is an on-chain account book that ensures that important agent behaviors follow specific constraints/rules.

The evolution of social norms: We will soon struggle with the manner and etiquette of interacting with artificial intelligence - we will say "please" and " Thank you"? Do we hold them morally responsible for their mistakes, or should we blame their developers?

I hope the marketing team thinks of this first because agencies are great at generating content and can live stream and post content on social media 24/7. If you are building an AI protocol, why not deploy a test agent on-premises to demonstrate performance?

In 2025, the lines between "employees" and "software" will begin to blur.

7. The ones who survive will be the truly useful 1%

We will see AI Darwinian evolution among agents. Why? Because running an AI agent requires computing power (i.e., reasoning cost), it is very expensive. If an agent cannot create enough value to pay its "rent," the game is over.

An example of an agent survival game is as follows:

Carbon Credit AI: Imagine an agent Decentralized energy networks can be searched, inefficiencies identified, and tokenized carbon credits traded autonomously. It makes enough money to pay its own computing bills. Then the agent will thrive.

DEX Arbitrage Bots: Agents that exploit price differences between decentralized exchanges can generate consistent revenue that covers their inference bills.

Shitposter on X: Meanwhile, what happens to virtual AI influencers with cute jokes but no sustainable revenue stream? Once the novelty wears off, token pricesIf the price plummets, it will fall into the quagmire and be unable to move forward.

The difference is clear: utility-driven agents thrive, while attention-fading agents fade into irrelevance.

This natural selection is beneficial to the development of this field. Developers are forced to innovate and prioritize production use cases over gimmicks. As these more powerful and effective agents emerge, they will eventually silence the skeptics (yes, even Kyle Samani).

8. Synthetic data will surpass human data

People say: "Data is the new oil." Artificial intelligence thrives on data, but its high data requirements are causing Concerns about impending data depletion.

Conventional wisdom holds that we should try to collect real-world private data from users, and even pay for it. But I’ve come to realize that a more practical path—especially in heavily regulated industries or where real data is scarce—lies in synthetic data.

These are human-generated datasets designed to simulate real-world data distributions. Providing scalable, ethical, privacy-friendly alternatives to human data.

Why synthetic data is effective:

Infinite scale: millions of medical records required X-rays or 3D scans of the factory? Synthetic generation can produce them in unlimited quantities, without waiting for real patients or real factories.

Privacy-friendly: There is no risk to personal information when processing human-generated data sets.

Customizable: You can tailor the distribution to your exact training needs, inserting edge cases that may be too rare or ethically complex to collect in the outside real world .

Yes, user-owned human data will still be important in many cases, but if synthetic data continues to improve in reality, it will likely increase in volume, generation Go beyond user data in terms of speed and freedom from privacy restrictions.

The next wave of decentralized artificial intelligence may center on "mini labs" that create highly specialized synthetic data sets for specific use cases.

These mini-labs will cleverly overcome regulatory and regulatory hurdles in data generation - just like Grass bypassed web scraping limitations by leveraging millions of distributed nodes.

9. Decentralization Transformed training will become truly useful

2024, Prime Intellect and Nous Pioneers like Research are pushing the boundaries of decentralized training. We have trained a 15 billion parameter model in a low-bandwidth environment, proving that large-scale training is possible outside of traditional centralized environments. p>

While these models aren't actually very useful compared to the existing base models (the performance is lower, so there isn't much reason to use them), I believe this This situation will change in 2025.

This week, EXO Labs goes one step further with SPARTA, reducing inter-GPU communication by more than 1,000 times. SPARTA enables large model training in slow-bandwidth environments without specialized infrastructure.

< p style="text-align: left;">What impressed me most was their statement: "SPARTA can work independently, but can also be combined with synchronous low-communication training algorithms (such as DiLoCo) for better performance . ”

This means that these improvements can be greatly improved when combined.

As With advances in techniques like model distillation, small models are becoming more useful and efficient, and the future of AI is not about size but about getting better and more accessible. Soon, we will have the ability to do it on edge devices and even mobile phones. High-performance models running on

10. There will be 10 new Crypto. AI protocol circulation valuation reaches US$1 billion

Welcome to the real gold rush

Many compare Virtuals and ai16z to early smartphones (iOS and Android), so it's easy to assume that the current leaders will continue to dominate.

But this market is too big and has not yet been fully developed. Just two companies areunable to dominate the market. By the end of 2025, I predict that at least 10 new Crypto AI protocols (that have not yet launched a token) will have a circulating (not fully diluted) market cap of over $1 billion.

Decentralized artificial intelligence is still in its infancy and will form a huge talent pool.

We are looking forward to the arrival of new protocols, new token models and new open source frameworks. These new competitors can displace existing competitors through a combination of incentives (such as airdrops or smart staking), technological breakthroughs (such as low-latency inference or chain interoperability), and user experience improvements (no code). Changes in public perception can be rapid and dramatic.

This is both the beauty and challenge of this field. Market size is a double-edged sword: the pie is huge, but the barrier to entry for technical teams is low. This laid the foundation for a Cambrian explosion of projects, with many projects gradually disappearing from sight and only a few becoming a force for change.

Bittensor, Virtuals and ai16z will not be alone for long. The next billion-dollar Crypto AI protocol is coming. For savvy investors, opportunities abound, which is why it's so exciting.

11. AI Agent: New Wave of App

When Apple launched the App Store in 2008, its slogan was "There's an app for that." To the effect: There is an app for all your needs)

Soon, you will say, "There's an agent for that.” (To the effect: All your needs can be solved by an AI agent)

You no longer need to click on the icon to open the app, but assign tasks to Delegate to a dedicated AI agent. These agents are context-aware, can cross-communicate with other agents and services, and can even automatically initiate tasks you never explicitly requested, such as monitoring your budget or re-planning your trip when flights change.

Simply put, your smartphone home screen may become a network of "digital colleagues", each with their own area of ​​expertise - health , finance, productivity and social.

Since these are encryptedAgents that can autonomously handle payments, authentication, or data storage using decentralized infrastructure.

12. Also, robots

Although most of the content of this article focuses on software , but I’m also very excited about the physical form of these AI revolutions: robots. Robotics will usher in its chatGPT era this decade.

The field still faces significant obstacles, particularly in accessing perception-based real-world datasets and improving physics performance. Some teams are tackling these challenges head-on, using cryptographic tokens to incentivize data collection and innovation. These efforts deserve attention.

I've been working in the tech industry for over a decade, and I can't remember the last time I felt this kind of genuine, visceral excitement. This wave of innovation feels different—bigger, bolder, and just getting started.