Top 10 Predictions for Crypto AI in 2025: Total Market Cap to Reach $150 Billion; 99% of AI Agents Will Vanish

By: blockbeats|2025/01/10 13:15:03
0
Share
copy
Original Article Title: What I'm Watching in 2025
Original Article Author: Teng Yan, Crypto Researcher
Original Article Translation: Felix, PANews

With the explosion of the AI industry this year, Crypto x AI has quickly emerged, and researcher Teng Yan, focusing on Crypto x AI, has published a post predicting 10 trends for 2025. Below are the details of the predictions.

1. The total market capitalization of Crypto AI tokens will reach $1.5 trillion

Top 10 Predictions for Crypto AI in 2025: Total Market Cap to Reach src=

Currently, the market value of Crypto AI tokens accounts for only 2.9% of the meme coin market value, but this ratio will not last long.

AI encompasses everything from smart contract platforms to memes, DePINs, and Agent platforms, data networks, and intelligent coordination layers. Its market presence alongside DeFi and memes is undeniable.

Why am I so confident about this?

· Crypto AI is a fusion of the two most powerful technologies

· AI Frenzy Trigger Event: An OpenAI IPO or similar event could trigger a global frenzy for AI. Meanwhile, Web2 capital has already started paying attention to decentralized AI infrastructure.

· Retail Frenzy: The AI concept is easy to understand and exciting, and retail investors can now invest in it through tokens. Remember the meme gold rush of 2024? AI will be a similar frenzy, just that AI is truly changing the world.

2. Bittensor Renaissance

The decentralized AI infrastructure Bittensor (TAO) has been online for many years as an established project in the Crypto AI space. Despite the AI craze, its token price has remained at levels from a year ago.

Today, Bittensor's Digital Hivemind has quietly achieved a leap: the registration fees for more subnets are lower, the performance of subnets in practical metrics such as inference speed surpasses that of Web2 peers, and EVM compatibility is introducing DeFi-like features into the Bittensor network.

Why Hasn't the TAO Token Surged? The drastic inflation plan and the market's focus on the Agent platform have hindered its rise. However, the upcoming dTAO (expected in the first quarter of 2025) could be a significant turning point. With dTAO, each subnet will have its own token, and the relative prices of these tokens will determine how emissions are distributed.

Why Bittensor Could Make a Comeback:

· Market-Based Emissions: dTAO will directly link block rewards to innovation and tangible performance. The better the subnet, the more valuable its token.

· Concentrated Capital Flow: Investors can ultimately target specific subnets they believe in. If a particular subnet wins out with an innovative distributed training approach, investors can deploy capital to represent their viewpoint.

· EVM Integration: EVM compatibility has attracted a broader crypto-native developer community within Bittensor, bridging the gap with other networks.

3. The Compute Market Is the Next "L1 Market"

The current unmistakable trend is the endless demand for compute.

NVIDIA CEO Jensen Huang has said that the demand for inference will grow "by a billion times." This exponential growth will disrupt traditional infrastructure plans, and new solutions are urgently needed.

A decentralized compute layer provides raw compute in a verifiable and economically efficient manner (for training and inference). Startups like Spheron, Gensyn, Atoma, and Kuzco are quietly building a solid foundation, focusing on product over tokens (these companies do not have tokens). As decentralized AI model training becomes practical, the entire potential market will see a sharp rise.

Comparison to L1:

· Similar to 2021: Remember Solana, Terra/Luna, and Avalanche competing for the "best" L1? There will be a similar competition among compute protocols, vying for developers and AI applications built on their compute layers.

· Web2 Demand: The $680 billion to $2.5 trillion cloud computing market dwarfs the crypto AI market. If these decentralized compute solutions can attract even a small portion of traditional cloud customers, the next wave of growth could see a 10x or 100x increase.

Just as Solana succeeded in the L1 space, the winner will dominate a whole new field. Pay close attention to reliability (such as robust service level agreements or SLAs), cost-effectiveness, and developer-friendly tools.

4. AI Agents Will Dominate Blockchain Transactions

Agent transaction on Gnosis' Olas; Source: Dune

By the end of 2025, 90% of on-chain transactions will no longer be initiated by real humans clicking "send" but rather by a cohort of AI agents carrying out tasks such as rebalancing liquidity pools, distributing rewards, or executing micro-payments based on real-time data feedback.

It might not sound far-fetched. Everything built over the past seven years (L1, rollups, DeFi, NFTs) has quietly paved the way for a world where AI operates on-chain.

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

Why this shift?

· Elimination of human error: Smart contracts execute exactly as coded. In turn, AI agents can process vast amounts of data faster and more accurately than real humans.

· Micro-payments: Transactions driven by these agents will become smaller, more frequent, and more efficient, especially as transaction costs trend downward on Solana, Base, and other L1/L2 solutions.

· Invisible infrastructure: If it removes some of the hassle, humans are more than happy to relinquish direct control.

AI agents will generate a significant amount of on-chain activity, so it's no surprise that all L1/L2 solutions are embracing agents.

The biggest challenge will be making these agent-run systems accountable to humans. As the ratio of agent-initiated transactions to human-initiated transactions continues to rise, new governance mechanisms, analytics platforms, and auditing tools will be necessary.

5. Interactions Between Agents: The Rise of Clusters

Source: FXN World

The concept of agent clusters—micro AI agents seamlessly collaborating to execute grand schemes—sounds like the plot of the next blockbuster sci-fi/horror movie.

Today's AI agents are mostly "lone wolves," operating in isolation with minimal and unpredictable interactions.

The Agent Cluster will change this situation, allowing AI agents' networks to exchange information, negotiate, and collaborate on decisions. Think of it as a decentralized collection of expert models, with each model contributing unique expertise to larger, more complex tasks.

One cluster might coordinate distributed computing resources on platforms like Bittensor. Another cluster could handle misinformation, validating the source in real-time before content spreads to social media. Each Agent in the cluster is an expert capable of precisely carrying out its task.

These cluster networks will yield intelligence stronger than any single isolated AI.

For clusters to thrive, universal communication standards are crucial. Regardless of their underlying framework, Agents need to discover, validate, and collaborate. Teams like Story Protocol, FXN, Zerebro, and ai16z/ELIZA are laying the groundwork for the emergence of Agent Clusters.

This underscores the pivotal role of decentralization. Task allocation across clusters in a transparent, on-chain rule management fashion makes the system more resilient and adaptive. If one agent fails, others step in.

6. Crypto AI Working Teams Will Be Human-Machine Hybrids

Source: @whip_queen_

Story Protocol hired Luna (an AI Agent) as its social media intern and pays her $1000 daily. Luna doesn't get along well with her human colleagues—she almost fired one while boasting about her stellar performance.

Though it sounds strange, this is a precursor to future AI Agents becoming true collaborators who have autonomy, accountability, and even wages. Companies across industries are beta-testing human-machine hybrid teams.

In the future, working with AI Agents will not be as slaves but as equal beings:

· Surge in productivity: Agents can handle vast amounts of data, communicate with each other, and make decisions around the clock without needing sleep or coffee breaks.

· Building trust through smart contracts: The blockchain is an impartial, tireless, and infinitely rememberable overseer. A ledger on the chain can ensure critical Agent operations follow specific boundary conditions/rules.

· Evolving Social Norms: Soon, we will start pondering the etiquette of interacting with an Agent—will we say "please" and "thank you" to AI? Will we hold them morally responsible for mistakes, or blame their developers?

The boundary between "employee" and "software" will begin to blur in 2025.

7. 99% of AI Agents Will Perish—Only the Useful Will Thrive

The future will witness a "Darwinian" elimination among AI agents. This is because running AI agents incurs a cost in terms of computational power (i.e., cost of reasoning). If an Agent cannot generate enough value to pay its "rent," the game is over.

Examples of Agent Survival Games:

· Carbon Credit AI: Imagine an Agent scouring a decentralized energy grid, identifying inefficiencies, and autonomously trading tokenized carbon credits. It thrives only if the money it earns is sufficient to cover its computational costs.

· DEX Arbitrage Bot: An Agent exploiting price differences between decentralized exchanges can earn a stable income, covering its reasoning costs.

· X Shitposter: A virtual AI KOL has amusing jokes, but lacks a sustainable income source? Once the novelty wears off (token price crashes), it cannot cover its costs.

Utility-driven Agents thrive, while attention-diverting Agents gradually become irrelevant.

This elimination mechanism benefits the industry. Developers are compelled to innovate, prioritizing use cases over gimmicks. With these more powerful, more efficient Agents emerging, they can silence the skeptics.

8. Synthetic Data Surpasses Human Data

"Data is the new oil." AI thrives on data, but its appetite has raised concerns about imminent data depletion.

The traditional view was to exhaustively gather users' private real data—even paying for it if necessary. However, a more practical approach is to use synthetic data, especially in heavily regulated industries or those with scarce real data.

Synthetic data is artificially generated datasets designed to mimic real-world data distributions. It provides a scalable, ethical, and privacy-friendly alternative to human data.

Why Synthetic Data Works So Well:

· Infinite Scale: Need a million medical X-rays or a 3D scan of a factory? Synthetic generation can produce an unlimited amount without waiting for real patients or real factories.

· Privacy-Friendly: When using artificially generated datasets, no personal information is at risk.

· Customizable: The distribution can be customized based on exact training needs.

Human-owned data remains crucial in many cases, but if synthetic data continues to improve in the real world, it may surpass user data in terms of quantity, generation speed, and privacy constraints.

The next wave of decentralized AI may center around "micro-labs" that can create highly specialized synthetic datasets tailored to specific use cases.

These micro-labs will cleverly sidestep policy and regulatory barriers in data generation—much like Grass circumvents network capture limits by leveraging millions of distributed nodes.

9. Decentralized Training Becomes More Practical

By 2024, pioneers such as Prime Intellect and Nous Research have pushed the boundaries of decentralized training. They trained a 150 billion-parameter model in a low-bandwidth environment, proving that large-scale training can occur beyond traditional centralized settings.

While these models were not practically useful compared to existing baseline models (low performance), this scenario is set to change in 2025.

This week, EXO Labs made further strides with SPARTA, reducing GPU-to-GPU communication by over 1,000 times. SPARTA enables large model training on slow bandwidth without the need for specialized infrastructure.

Impressively, their statement reads, "SPARTA can operate standalone or synergistically with synchronous low-communication training algorithms like DiLoCo for enhanced performance."

This means these improvements can stack, thereby increasing efficiency.

As technology progresses, micro-models become more practical and efficient, signaling that the future of AI lies not in scale but in becoming better and more user-friendly. Soon, there are expectations of having high-performance models runnable on edge devices or even smartphones.

10. Ten New Crypto AI Protocol Tokens Reach a $10 Billion Circulating Market Cap (Yet to Launch)

a16z to Hit $20 Billion Valuation in 2024

Welcome to the real gold rush.

It's easy for people to assume that current leaders will continue to dominate, with many drawing comparisons between Virtuals and a16z to early days of smartphones (iOS and Android).

However, this market is too vast and underdeveloped to be ruled by just two players. By the end of 2025, it is projected that at least ten new encrypted AI protocols (with tokens not yet released) will have a circulating (undiluted) market cap of over $1 billion.

Decentralized AI is still in its infancy. Additionally, the pool of talent is continuously growing.

Expect the arrival of new protocols, novel token models, and new open-source frameworks. These new entrants may displace existing players through a combination of incentive mechanisms (such as airdrops or clever staking), technical breakthroughs (like low-latency inference or chain interoperability), and user experience enhancements (no code). The shift in public perception could be sudden and dramatic.

This is both the beauty and the challenge of this field. Market size is a double-edged sword: the cake is huge, but the barrier to entry for tech teams is low. This sets the stage for explosive growth in projects, with many fading away gradually, but a few holding transformative power.

Bittensor, Virtuals, and a16z will not hold the lead forever. The next billion-dollar encrypted AI protocol is on the horizon. Savvy investors have plenty of opportunities, which is what makes it so exciting.

Original Article Link

You may also like

a16z Leads $18M Seed Round for Catena Labs, Crypto Industry Bets on Stablecoin AI Payment

Traditional finance is still stuck in a "human-to-human" model, while Catena aims to achieve "AI-to-AI" interaction.

Pharos, deeply integrated with AntChain, is about to launch. How can we get involved?

What is the relationship between the $8 million funded NewChain and Ant, and how will they interact?

$COIN Joins S&P 500, but Coinbase Isn't Celebrating

On May 13, S&P Dow Jones Indices announced that Coinbase would officially replace Discover Financial Services in the S&P 500 on May 19. While other companies like Block and MicroStrategy, closely tied to Bitcoin, were already part of the S&P 500, Coinbase became the first cryptocurrency exchange whose primary business is in the index. This also signifies that cryptocurrency is gradually moving from the fringes to the mainstream in the U.S.



On the day of the announcement, Coinbase's stock price surged by 23%, surpassing the $250 mark. However, just 3 days later, Coinbase was hit by two consecutive events: a hack where employees were bribed to steal customer data and a demand for a $20 million ransom, and an investigation by the U.S. Securities and Exchange Commission (SEC) into the authenticity of its claim of having over 100 million "verified users" in its securities filings and marketing materials. These two events acted as mini-bombs, and at the time of writing, Coinbase's stock had already dropped by over 7.3%.


Coincidentally, Discover Financial Services, being replaced by Coinbase, can also be considered the "Coinbase" of the previous payment era. Discover is a U.S.-based digital banking and payment services company headquartered in Illinois, founded in 1960. Its payment network, Discover Network, is the fourth largest payment network apart from Visa, Mastercard, and American Express.


In April, after the approval of the acquisition of Discover by the sixth-largest U.S. bank, Capital One, this well-established digital banking company of over 60 years smoothly handed over its S&P 500 "seat" to this emerging cryptocurrency "bank." This unexpected coincidence also portrayed the handover between the new and old eras in Coinbase's entry into the S&P 500, resembling a relay race scene. However, this relay baton also brought Coinbase's accumulated "external troubles and internal strife" to a tipping point.


Side Effects of ETFs


Over the past decade, cryptocurrency exchanges have been the most stable "profit machines." They play a role in providing liquidity to the entire industry and rely on trading fees to sustain their operations. However, with the comprehensive rollout of ETF products in the U.S. market, this profit model is facing unprecedented challenges. As the leader in the "American stack," with over 80% of its business coming from the U.S., Coinbase is most affected by this.



Starting from the approval of Bitcoin and Ethereum spot ETFs, traditional financial capital has significantly onboarded users and funds that originally belonged to exchanges in a more cost-effective, compliant, and transparent manner. The transaction fee revenue of cryptocurrency exchanges has started to decline, and this trend may further intensify in the coming months.


According to Coinbase's 2024 Q4 financial report, the platform's total trading revenue was $417 million, a 45% year-on-year decrease. The contribution of BTC and ETH's trading revenue dropped from 65% in the same period last year to less than 50%.


This decline is not a result of a decrease in market enthusiasm. In fact, since the approval of the Bitcoin ETF in January 2024, the inflow of BTC into the U.S. market has continued to reach new highs, with asset management giants like BlackRock and Fidelity rapidly expanding their management scale. Data shows that BlackRock's iShares Bitcoin ETF (IBIT) alone has surpassed $17 billion in assets under management. As of mid-May 2025, the cumulative net inflow of 11 major institutional Bitcoin spot ETFs on the market has exceeded $41.5 billion, with a total net asset value of $1214.69 billion, accounting for approximately 5.91% of the total Bitcoin market capitalization.


Chart showing the trend of net outflows for Grayscale among the 11 institutions


Institutional investors and some retail investors are shifting towards ETF products, partly due to compliance and tax considerations. On one hand, ETFs have much lower trading costs compared to cryptocurrency exchanges. While Coinbase's spot trading fee rate varies annually in a tiered manner but averages around 1.49%, for example, the management fee for IBIT ETF is only 0.25%, and the majority of ETF institution fees fluctuate around 0.15% to 0.25%.



In other words, the more rational users are, the more likely they are to move from exchanges to ETF products, especially for investors aiming for long-term holdings.


According to multiple sources, several institutions, including VanEck and Grayscale, have submitted applications to the SEC for a Solana (SOL) ETF, with some institutions also planning to submit an XRP ETF proposal. Once approved, this may trigger a new round of fund migration. According to a report submitted by Coinbase to the SEC, as of April, the platform's trading revenue from XRP and Solana accounted for 18% and 10%, nearly one-third of the platform's fee revenue.



However, the Bitcoin and Ethereum ETFs passed in 2024 also reduced the fees for these two tokens on Coinbase from 30% and 15% to 26% and 10%, respectively. If the SOL and XRP ETFs are approved, it will further undermine the core fee revenue of exchanges like Coinbase.


The expansion of ETF products is gradually weakening the financial intermediary status of cryptocurrency exchanges. From their original roles as matchmakers and clearers to now gradually becoming mere "on-ramps and off-ramps" for funds, exchanges are seeing their marginal value squeezed by ETFs.


Robinhood Takes a Stand, Traditional Brokerages Join the Fray


On May 12, 2025, SEC Chairman Paul S. Atkins gave a keynote speech at the Tokenization and Cryptocurrency Working Group roundtable. The theme of his speech revolved around "It is a new day at the SEC," where he indicated that the SEC would not approach enforcement and regulation the same way as before but would instead pave the way for cryptocurrency assets in the U.S. market.



With signs of cryptocurrency compliance such as the SEC's "NEW DAY" declaration, an increasing number of traditional brokerages are attempting to enter the cryptocurrency industry. One of the most representative cases is the well-known U.S. brokerage Robinhood, which began expanding its crypto business in 2018. By the time of its IPO in 2021, Robinhood's crypto business revenue accounted for over 50% of the company, with a significant boost from the Dogecoin "moonshot" promoted by Musk.


In Q1 2025 earnings report, Robinhood showcased strong growth, especially in revenue from cryptocurrency and options trading. Fueled by Trump's Memecoin, cryptocurrency-related revenue reached $250 million, nearly doubling year-over-year. Consequently, Robinhood Gold subscription users reached 3.5 million, a 90% increase from the previous year, with the rapid growth of Robinhood Gold providing the company with a stable source of income.



Meanwhile, RobinHood is actively pursuing acquisitions in the cryptocurrency space. In 2024, it announced a $2 billion acquisition of the long-standing European cryptocurrency exchange Bitstamp. Additionally, Canada's largest cryptocurrency CEX, WonderFi, which recently went public on the Toronto Stock Exchange, also announced its integration with RobinHood Crypto. After obtaining virtual asset licenses in the UK, Canada, Singapore, and other markets, RobinHood has taken a proactive approach in the compliant cryptocurrency trading market.



Furthermore, an increasing number of brokerage firms are exploring the same path. Futu Securities, Tiger Brokers, and others are also dipping their toes into cryptocurrency trading, with some having applied for or obtained the VA license from the Hong Kong SFC. Although their user bases are currently small, traditional brokerages have a natural advantage in user trust, regulatory licenses, and low fee structures. This could pose a threat to native cryptocurrency platforms in the future.



User Data Breach: Is Coinbase Still Secure?


In April 2025, security researchers discovered that some Coinbase user data was leaked on the dark web. While the platform initially responded by attributing it to a "technical misinformation," it still raised concerns among users regarding its security and privacy protection. Just two days before Dow Jones Indexes announced Coinbase's addition to the S&P 500 Index, on May 11, 2025, Coinbase received an email from an unknown threat actor claiming to have obtained customer account information and internal documents, demanding a $20 million ransom to keep the data private. Subsequent investigations confirmed the data breach.


Cybercriminals obtained the data by bribing overseas customer service agents and support staff, mainly in "non-U.S. regions such as India." These agents abused their access to Coinbase's internal customer support system and stole customer data. As early as February this year, blockchain detective ZachXBT revealed on X platform that between December 2024 and January 2025, Coinbase users lost over $65 million to social engineering scams, with the actual amount potentially higher.


Among the victims was a well-known figure, 67-year-old Ed Suman, an established artist in the art world for nearly two decades, having been involved in the creation of artworks such as Jeff Koons' "Balloon Dog" sculpture. Earlier this year, he fell victim to an impersonation scam involving fake Coinbase customer support, resulting in a loss of over $2 million in cryptocurrency. ZachXBT critiqued Coinbase for its inadequate handling of such scams, noting that other major exchanges have not faced similar issues and recommending Coinbase to enhance its security measures.


Amidst a series of ongoing social engineering incidents, although there has not been any impact on user assets at the technical level so far, it has raised concerns among many retail and institutional investors. Especially institutions holding massive assets on Coinbase. Just considering the U.S. BTC ETF institutions, as of mid-May 2025, they collectively hold nearly 840,000 BTC, and 75% of these are custodied by Coinbase. If we price BTC at $100,000, this amount reaches a staggering $63 billion, which is equivalent to the nominal GDP of two Iceland in the year 2024.


Visualization: ChatGPT, Source: Farside


In addition, Coinbase Custody also serves over 300 institutional clients, including hedge funds, family offices, pension funds, and endowments. As of the Q1 2025 financial report, Coinbase's total assets under management (including institutional and retail clients) reached $404 billion. The specific amount of institutional custodied assets was not explicitly disclosed in the latest report, but it should still be over 50% based on the Q4 2024 report.


Visualization: ChatGPT


Once this security barrier is breached, not only could the rate of user attrition far exceed expectations, but more importantly, institutional trust in it would undermine the foundation of its business. Therefore, after a hacking event, Coinbase's stock price plummeted significantly.


CEXs are All in Self-Rescue Mode


Facing a decline in spot trading fee revenue, Coinbase is also accelerating its transformation, attempting to find growth opportunities in derivatives and emerging assets. Coinbase acquired a stake in the options platform Deribit at the end of 2024 and announced the official launch of perpetual contract products in 2025. This acquisition fills in Coinbase's gap in options trading and its relatively small global market share.



Deribit has a strong presence in non-U.S. markets, especially in Asia and Europe. The acquisition has enabled Coinbase to gain a dominant position in bitcoin and ethereum options trading on Deribit, accounting for approximately 80% of the global options trading volume, with daily trading volume remaining above $2 billion.


Meanwhile, 80-90% of Deribit's customer base consists of institutional investors, with their professionalism and liquidity in the Bitcoin and Ethereum options market highly favored by institutions. Coinbase's compliance advantage, coupled with its already robust institutional ecosystem, makes it even more suitable. By using institutions as an entry point, it can face the squeeze from giants like Binance and OKX in the derivatives market.



Facing a similar dilemma is Kraken, which is attempting to replicate Binance Futures' model in non-U.S. markets. Since the derivatives market relies more on professional users, fee rates are relatively higher and stickiness is stronger, making it a significant source of revenue for exchanges. In the first half of 2025, Kraken completed the acquisition of TradeStation Crypto and a futures exchange, aiming to build a complete derivatives trading ecosystem to hedge the risk of declining spot transaction fee income.


With the surge of Memecoin in 2024, Binance, OKX, and various CEX platforms began massively listing small-market-cap, highly volatile tokens to activate active trading users. Due to the wealth effect and trading activity of Memecoins, Coinbase was also forced to join the battle, successively listing popular tokens from the Solana ecosystem such as BOOK OF MEME and Dogwifhat. Although these coins are controversial, they are frequently traded, with fee rates several times higher than mainstream coins, serving as a "blood-boosting" method for spot trading.


However, due to its status as a publicly traded company, this practice is a riskier endeavor for Coinbase. Even in the current crypto-friendly environment, the SEC is still investigating whether tokens like SOL, ADA, and SAND constitute securities.


In addition to the forced transformation strategies carried out by the aforementioned CEXs, they are also starting to lay out RWAs and the most talked-about stablecoin payment fields, such as the PYUSD launched through a collaboration between Coinbase and Paypal, Coinbase's support for the Euro stablecoin EURC by Circle that complies with EU MiCA regulatory requirements, or the USD1 launched through a collaboration between Binance and WIFL. In the increasingly crowded trading field, many CEXs have shifted their focus from just the trading market to the application field.


The golden age of transaction fees has quietly ended, and the second half of the crypto exchange platform game has silently begun.


Key Market Insights for May 16th, how much did you miss out on?

1. On-chain Flows: $111.3M inflow to Ethereum this week; $237.6M outflow from Berachain 2. Largest Price Swings: $ETHFI, $NEIRO 3. Top News: Data: Solana Network's revenue reached $7.9M on the 13th, surpassing the sum of all other L1 and L2 chains

MOG Coin Skyrockets as Elon Musk and Garry Tan Embrace "mog/acc" Identity

「mog/acc」 is rapidly sweeping through various figures, from Elon Musk to Garry Tan, boosting the project's visibility and ultimately driving up the price.

The End and Rebirth of NFTs: How the Meme Coin Craze Ended the PFP Era?

There must be another Labubu hidden beneath the ruins.

Popular coins

Latest Crypto News

Read more