From the resurgence of Bittensor to the rise of AI Agents, the top ten predictions for crypto AI in 2025

Title: What We're Watching in 2025 (Crypto AI)

Author: Teng Yan

Compiled by: Asher

The Crypto AI sector is full of appeal, and although it lacks historical precedents and clear trends, this also means it is at a brand-new starting point, waiting for future development. Thinking about reviewing all this by 2026, seeing the gap between the expectations at the beginning of 2025 and the actual situation, will be even more exciting.

I. The total market capitalization of the Crypto AI sector will reach $150 billion

Currently, the tokens in the Crypto AI sector only account for 2.9% of the Altcoin market value, but this proportion is unlikely to last too long. As artificial intelligence gradually expands into new areas such as smart contract platforms, Meme, decentralized physical infrastructure (DePIN), agent platforms, data networks, and smart coordination layers, its integration with DeFi and Meme tokens has become an inevitable trend.

The confidence in the Crypto AI sector is because it is at the intersection of two of the most powerful technological trends:

• AI Hype Trigger Events: The IPO of OpenAI or similar events may trigger a global frenzy for AI. At the same time, Web2 institutional capital has already been paying attention to decentralized AI infrastructure as an investment.
• Retail Frenzy: The concept of artificial intelligence is easy to understand and exciting, and now they can invest in it through tokens. Remember the Meme coin gold rush in 2024? This will be the same frenzy, only this time artificial intelligence will change the world more tangibly.

II. The Revival of Bittensor

Bittensor (token name TAO) has been around for years. It is an old player in this field. Despite the hype around artificial intelligence, its token price has been hovering around the level of a year ago. However, the digital hive mind behind Bittensor has been quietly advancing, with more subnets emerging, registration fees decreasing, and the subnets outperforming the corresponding Web2 technologies in terms of inference speed and other practical performance, while the introduction of EVM compatibility has also brought in DeFi-like functionalities, further enriching Bittensor's network.

So why hasn't TAO skyrocketed? The steep inflationary schedule and the shift of attention to platforms oriented towards AI Agents have limited it. However, dTAO (expected in Q1 2025) may be a major turning point. Through dTAO, each subnet will have its own token, and the relative prices of these tokens will determine how the releases are allocated.

Why Bittensor has a chance to rekindle:

• Market-based Releases: dTAO will directly link block rewards to innovation and measurable performance. The better a subnet performs, the more valuable its token will be - and thus the more it will receive in releases.
• Focused Capital Flows: Investors will finally be able to invest in specific subnets they are bullish on. If a subnet adopts innovative approaches to distributed training and achieves success, capital can flow into that subnet to express the investment view.
• EVM Integration: EVM compatibility has attracted a wider crypto-native developer community to Bittensor, bridging the gap with other networks.

We are currently tracking the various subnets and recording their actual progress in their respective fields. At some point, a DeFi summer-like event for @opentensor is expected.

III. The Computing Market is the Next L1 Battleground

The insatiable demand for computing will become an obvious mega-trend. Nvidia CEO Jensen Huang has stated that inference demand will increase "a billion-fold". This exponential growth will break the planning of traditional infrastructure and urgently call for "new solutions".

The decentralized computing layer provides raw computing (for training and inference) in a verifiable and cost-effective way. Startups like @spheronfdn, @gensynai, @atoma_network, and @kuzco_xyz are quietly building strong foundations to leverage this, focusing on products rather than tokens (these companies currently have no tokens). As decentralized AI model training becomes feasible, the addressable market is expected to skyrocket.

Comparing the Crypto AI sector to the L1 sector:

• Just like 2021: Remember how Solana, Terra, and Avalanche fought to be the "best" L1? We will see similar battles between computing protocols, competing for developers and AI applications to build on their computing layers.
• Web2 Demand: The $680 billion to $2.5 trillion cloud computing market far exceeds the Crypto AI market. If these decentralized computing solutions can capture even a small portion of traditional cloud customers, we could see the next 10x or 100x growth wave.

Just as Solana stood out in the L1 space, the winner here will dominate a brand-new frontier, and the three key criteria to watch are reliability, cost-effectiveness, and developer-friendly tools.

IV. AI Agents Will Flood Blockchain Transactions

By the end of 2025, 90% of on-chain transactions will no longer be manually triggered by humans clicking "send". Instead, these transactions will be executed by an army of AI Agents, continuously rebalancing liquidity pools, allocating rewards, or executing micro-payments based on real-time data sources.

This is not as far-fetched as it may seem. Everything we have built over the past seven years (L1, rollups, DeFi, NFT, etc.) has quietly paved the way for an AI-driven on-chain world.

@autonolas agents transacting on Gnosis

So, why this transformation?

• No More Human Errors: Smart contracts execute precisely as coded. AI Agents can process large amounts of data faster and more accurately than a group of humans.
• Micro-payments: AI Agent-driven transactions will become smaller, more frequent, and more efficient, especially as transaction costs on Solana, Base, and other L1/L2s trend downwards.
• Invisible Infrastructure: Humans will be happy to relinquish direct control if it means less hassle. Trusting Netflix to auto-renew, trusting an AI Agent to automatically rebalance a user's DeFi positions is the natural next step.

AI Agents will generate astonishing on-chain activity, but the biggest challenge will be making these AI Agent-driven systems accountable to humans. As the ratio of AI Agent-initiated transactions to human-initiated transactions increases, new governance mechanisms, analytics platforms, and auditing tools will be needed.

V. Interactions Between Agents: The Rise of the AI Collective Concept

AI Agent collectives refer to seamlessly collaborating micro-AI entities executing grand plans, which sounds like the plot of the next hit sci-fi or horror movie. Currently, most AI Agents operate in isolation, with little interaction and unpredictability. However, AI Agent collectives will change this, allowing multiple AI Agents to exchange information, negotiate, and make joint decisions across the network.

These AI Agent collectives can be viewed as decentralized professional model collectives, with each model contributing its unique expertise to larger, more complex tasks. The potential is staggering. For example, one collective may coordinate distributed computing resources on platforms like Bittensor, while another collective could verify content sources in real-time to prevent the spread of misinformation on social media. Each AI Agent in the collective is an expert, precisely executing its own task.

山寨币 networks' intelligence will far exceed that of any single isolated AI. Universal communication standards are crucial for agent collectives to thrive. Agents need to be able to discover, authenticate, and collaborate without being constrained by underlying frameworks. Teams like Story, FXN, ZEREBRO, and ai16z are working to lay the groundwork for the rise of agent collectives. At the same time, this also highlights the critical role of decentralization, allocating tasks to agent collectives governed by transparent on-chain rules, giving the system greater resilience and adaptability. If one agent fails, others can step in to fill the gap and maintain the system's continuous operation.

VI. Crypto AI teams will be human-AI hybrids

Story has hired Luna (an AI Agent project) as their social media intern, paying her $1,000 per day. It may sound strange, but this is a harbinger of the future, where AI Agents will become true collaborators, with their own autonomy, responsibilities, and even salaries. Across industries, companies are experimenting with human-machine hybrid teams. We will work hand-in-hand with AI Agents, not as our slaves, but as equal partners: - Productivity surge: AI Agents can process massive amounts of data, communicate with each other, and make decisions 24/7 without the need for sleep or coffee breaks. - Trust established through smart contracts: Blockchain is an impartial, tireless, and unforgetting overseer. An on-chain ledger ensures that critical AI Agent behaviors adhere to specific boundary conditions/rules. - Social norms are evolving: We will soon face etiquette around interacting with intelligent agents - should we say "please" and "thank you" to AI? Should we hold them morally responsible for their mistakes, or blame their developers? The line between "employee" and "software" will start to blur by 2025. Expect more crypto teams to get involved, as AI Agents excel at content generation, can operate 24/7, and publish on social media. If you're developing an AI protocol, why not showcase its capabilities by deploying internal AI Agents?

VII. 99% of AI Agents will perish (only the useful ones will survive)

We will see a Darwinian selection among AI Agents. This is because running an AI Agent requires computational power, i.e., inference costs. If an AI Agent cannot generate enough value to cover its "rent," it will face extinction. For example, in an AI Agent survival game, first, there's the carbon credit AI Agent: Suppose an AI Agent is scouring a decentralized energy grid for inefficiencies and autonomously trading tokenized carbon credits. If it can earn enough revenue to cover its inference costs, this AI Agent will thrive. Another example is the DEX arbitrage bot: This AI Agent earns stable income by exploiting price differences between decentralized exchanges, enough to cover its inference costs. In contrast, the X prankster: an entertaining but unsustainable virtual AI influencer, as the novelty wears off and token prices decline, it will gradually disappear, unable to sustain itself. The distinction is clear - AI Agents oriented towards utility will flourish, while those relying on disruption and gimmicks will become irrelevant. This natural selection benefits the industry, as it compels developers to continuously innovate, prioritizing productive applications over flashy technology. As more powerful and productive AI Agents rise, skeptics will gradually fall silent.

VIII. AI-generated synthetic data will surpass human data

The saying "data is the new oil" has become widespread. However, AI's heavy reliance on data has also raised concerns about an impending data shortage. The traditional view is that we should find ways to collect private real-world data from users, even paying them for it. However, in highly regulated industries or where real-world data is scarce, a more practical solution may be synthetic data. Synthetic data is artificially generated, designed to mimic the distribution of real-world data. It provides a scalable, ethically-friendly, and privacy-safe alternative to human data. The advantages of synthetic data include: - Unlimited scale: Whether it's a million medical X-rays or 3D scans of a factory, synthetic data can be generated in unlimited quantities, without relying on real patients or factories. - Privacy-friendly: When working with synthetic data, personal privacy information is not at risk. - Customizable: Data distributions can be adjusted to specific training needs, even inserting rare or ethically complex edge cases that are scarce in reality. While human-generated data will still be important in many cases, if synthetic data continues to improve in realism, it may surpass user data in terms of quantity, generation speed, and lack of privacy constraints. Future decentralized AI may revolve around "mini-labs" focused on creating highly specialized synthetic data sets to meet specific use cases.

IX. Decentralized training starts to pay off

In 2024, pioneers like Prime Intellect and Nous Research pushed the boundaries of decentralized training. For example, they successfully trained a 15 billion parameter model in low-bandwidth environments, proving that large-scale training can be achieved outside of traditional centralized setups. While these models may not perform as well as existing base models in practical applications, and thus have limited use cases, this is expected to change in 2025. EXO Labs further advanced the progress through SPARTA, reducing GPU-to-GPU communication by over 1000x. SPARTA makes large model training possible in low-bandwidth environments without relying on specialized infrastructure. Most impressively, they stated: "SPARTA works standalone, but can also be combined with synchronous low-communication training algorithms like DiLoCo for even better performance." This means these improvements are additive, with efficiency gains compounding over time. As model technologies continue to improve, smaller and more efficient models will become increasingly useful, and the future of AI will focus not just on scale, but on quality and accessibility. Soon, we will see high-performance models capable of running on edge devices, even smartphones.

X. At least ten new crypto AI superprotocols

While many compare Virtuals and ai16z to the early stages of smartphones (like iOS and Android), and believe the current leaders will continue to dominate, this market is vast and largely untapped - two participants cannot possibly control it. By the end of 2025, it is expected that at least ten new crypto AI protocols (not yet launched) will reach a market cap of over $1 billion. Decentralized AI is still in its infancy, and a wealth of talent is converging. New protocols, new token models, and new open-source frameworks will continue to emerge, and these new entrants may displace the incumbents through incentives (like airdrops or clever staking), technological breakthroughs (like low-latency inference or cross-chain interoperability), and user experience improvements (like no-code). Changes in public perception could be sudden and dramatic. Bittensor, Virtuals, and ai16z won't be alone for long, the next billion-dollar crypto AI protocol is coming, and savvy investors will face a wealth of opportunities - this is what makes this market so exciting.