DeAI remains one of the biggest opportunities in the next two years; the POW+AI project Pearl is being discussed; the next step for exchanges is AI-driven personalization for every individual; the era of traditional agencies making money off information asymmetry is coming to an end; but if you're not a practitioner and have no social needs, and are just trading coins or interested in technology, there's really no need to come offline.
Compiled by: Yuliya, PANews
At this Singapore TOKEN2049, many people's impressions were not entirely consistent: some felt there were many people, many events, and high social density; others felt that truly new projects had become fewer, Builders were fewer, and institutions, traders, KOLs, and funders were more numerous.
The industry no longer only looks at who tells a good narrative, but at who has real users, real revenue, real assets, and real liquidity. The main venue increasingly resembles a traditional financial expo, with stablecoins, payments, RWA, privacy, Perp, prediction markets, and others taking center stage, while AI/DeAI remains one of the biggest variables of concern in the Builder circle......
Below is a collection of TOKEN2049 observations from different attendees:
0xJeff: Fewer Builders, More Institutions, but DeAI Is Still One of the Biggest Opportunities in the Next Two Years
0xJeff used three days to record his observations of the changes in the atmosphere at TOKEN2049.
On the first day, he found that 90% of conversations revolved around: RWA, tokenized stocks, structured products (crypto institutionalization), while the other 10% involved AI, financialization of compute, and everything going on-chain.
He noticed that VCs, liquidity funds, and traders were all talking about Grass and their bullish stance on it, while Venice, Near, Pumpfun, and Orbio were also frequently mentioned.
Most people have not realized the ongoing on-chain/DeAI wave, and most people also do not understand the value proposition of on-chain/decentralized inference, x402, and the emerging tokenized inference credit market.
Many teams are building front ends, trading platforms, AI agent trading strategy platforms, and structured product protocols around platforms such as Hyperliquid, Ondo, and Polymarket. More people are going on-chain because the experience has improved, and tokenized assets offer more value than traditional brokerages, such as: 24/7 trading, asset collateralization, lending yield, basis trading, and more flexible portfolio strategies.
But overall, the crowd was down by more than 50% compared with before, with fewer Builders and more traders, investors, KOLs, and institutions. The overall atmosphere was more institutionalized than in the past, and retail excitement was much lower.
On the second day, this feeling was further confirmed. The main venue was dominated by exchanges, custody, payments, stablecoin infrastructure, and neobanks. Many investors have become pessimistic about the crypto market, either neutral or bearish, hoarding cash, and holding only 1 or 2 positions they believe in. Because they believe DeFi has matured, the winners have already emerged, and NFT and meme narratives are gradually becoming outdated.
Only a few narratives and teams in the market are ready for the next rally, such as: AI, privacy, RWA, payments.
However, some people are still quite bullish: Eastern whales are more pessimistic, Western whales are more optimistic.
On the third day, at the gm AI v2 event, he finally felt the Builder atmosphere of 2023 again. More geeky discussions around AI, privacy, robotics, DePIN, decentralization, tokenomics, and more reappeared.
However, many projects that have long been deeply involved in the industry are shifting toward new development directions:
Some projects have begun to position themselves in China's AI market and move their offices to Shenzhen;
Some have shifted toward combining AI with gaming and consumer content;
Others are focusing on social trading and privacy AI.
It is worth noting that although Twitter is abuzz with discussion about quantum computing, almost no one mentioned the topic on-site at Token2049.
As for the market, bears are mainly worried about the macro environment and the four-year cycle theory, while bulls firmly believe in the institutionalization of cryptocurrency and its potential to integrate with TradFi as a pillar of global finance.
Builders at this stage are more focused on the essence of business, with key discussions on how to increase user spending, how to effectively align interests and drive growth through token mechanisms, and how to use x402 to accelerate product launches. This shift from speculation to building fundamental products has brought a new atmosphere to the industry.
His final conclusion is: on-chain / decentralized AI is one of the most asymmetric and fastest-growing areas in Crypto over the next two years.
Although the main venue is becoming increasingly institutionalized, the truly interesting new research directions are still hidden in discussions among AI Builders.
Rocky: The Next Step for Exchanges Is AI-Driven Personalization at Scale
Trader Rocky attended a small Binance gathering and believed the event had a high density of information exchange, and he could also feel Binance's emphasis on the AI direction.
His core observation is: in the future, exchanges will increasingly move toward "personalization at scale."
Today, when everyone opens the same exchange, the interface and features they see are largely the same. But in the future, different users' needs will be understood more deeply by AI:
People who like contracts will see more contract-related data and strategies;
People who study options will see information more suitable for options decisions;
People focused on wealth management will see yield, risk, and asset allocation tools;
New users may receive simpler, lower-threshold guidance.
As AI increasingly understands user habits and preferences, the process of finding information, finding features, and making decisions will all become simpler. In the future, the key competitive focus of exchanges will not only be who has more coins, better liquidity, and lower fees, but who understands users better and who can truly build user needs into products.
Bennie: Fewer Projects, but Real Alpha Is Hidden Offline and in Long-Termism
Bennie's first impression was: Crypto projects are visibly decreasing.
From the offline feel, there are not as many new projects worth paying attention to as before. In the past, after everyone got tired of VC coins, they turned to Alpha shells, but as listing difficulty increases, the assembly-line model of forming groups, issuing tokens, and listing will become increasingly difficult. Conversely, some "shells" that themselves have resources, channels, and narrative-carrying capacity may rise in value in the future, and a market specifically for speculating on shells has even already appeared.
The direction he is most bullish on is RWA, especially U.S. equity-related infrastructure. In the past, RWA mostly remained at the level of tokenizing U.S. Treasuries, but with the development of U.S. equity Perps, 24/7 trading, and price discovery during non-traditional trading hours, the trend of traditional financial assets and trading infrastructure entering Crypto is becoming increasingly obvious. Platforms like Robinhood are bringing traditional financial users into Crypto, and Hyperliquid's HIP-3 has also made more people realize the potential of bringing traditional assets on-chain.
Another important judgment is: the era in which traditional Agencies make money from information asymmetry is coming to an end.
AI is rapidly lowering the cost of research, content production, and information acquisition. What used to require a team to accomplish can now be done by a few people working with AI. Agencies that simply sell manpower and resources will find it increasingly difficult, and in the future more attention should be paid to Advisor / CMO-type roles deeply involved in project growth, as well as new GTM models such as crowdsourced content, UGC, and short-video matrices.
In offline socializing, Bennie suggests paying more attention to the people at the dinner table who do not talk much.
Real Alpha is often not the person who is best at introducing themselves and packaging themselves, but those low-key people who truly have something. On the contrary, if a person keeps loudly talking about what they have done, it is most likely just BD.
He also encourages everyone to boldly pitch projects they like. Many Founding Teams are actually on-site, and collaboration opportunities are often created by simply walking over and talking.
Finally, he mentioned several very practical suggestions:
showoff > execute: everyone should manage their own social media. If you like thinking, write more Research and publish your views. If you are a Marketer, try more to show your differentiators through video, content, and different forms of expression.
Make good use of AI; most people have developed less than 1% of AI's potential.
Plan ahead for a second identity or overseas visa, because TOKEN2049 in 2027 will be held in New York, and Crypto's resources and liquidity are further moving toward the United States.
But if you are not a practitioner and have no social needs, and are just trading coins or interested in technology, then there is actually no need to come offline; asking AI directly may be more efficient.
Yue Xiaoyu: The Main Venue Is a Business Networking Event, the BNB Venue Is a Chinese Players' Garden Party
Yue Xiaoyu's observations mainly come from two scenarios: the TOKEN2049 main venue and the BNB Chain venue. Her most obvious feeling is: TOKEN2049 and BNB Chain are two separate circles.At TOKEN2049's main venue, about 80% of attendees are from overseas, and the booths are mostly occupied by exchanges and crypto payment projects, making the overall scene feel more like one for industry project teams, institutions, and business partnerships.
At the BNB Chain venue, by contrast, about 80% are Chinese-speaking attendees, and participants are mostly KOLs, traders, and whales. The projects on site are also more focused on the BNB ecosystem itself, including memes, stablecoin payments, on-chain contracts, prediction markets, and more.
If the TOKEN2049 main venue is a business networking event for industry projects, then the BNB Chain venue feels more like a carnival for Chinese-speaking players.
The proactive inquiries she received at the main venue mainly fell into three categories:
Crypto payment service providers: Fiat/Crypto two-way payment solutions have already become relatively mature. Overseas compliant on-ramp/off-ramp and localized payment infrastructure are becoming essential for projects expanding abroad.
Regional marketing agencies: They can handle market promotion in regions such as Europe and the United States, Latin America, and Africa. It is clear that if you want to go global, you need to shift toward refined, regionally localized operations. Different countries and regions vary greatly, and market expansion methods are also very different.
Other project teams and industry practitioners: This group may not necessarily convert directly, but it brings brand awareness and exposure within relevant circles.
Online communities are flooded with market noise and negative sentiment, and everyone is very anxious; but offline, many practitioners, users, and partners are still actively looking for reliable projects, building relationships, and driving cooperation. She therefore reached a judgment: online public opinion is dominated by emotion and market conditions, while offline events are dominated by value and networks.
As for whether project teams should exhibit, her advice is:
For projects that have just started, it is not recommended to rent a booth at the main venue, as ROI is low;
For projects in a high-growth phase or mature phase, a main venue booth is necessary, because it is not for direct conversion, but for showing strength as a brand and gaining industry exposure.
But whether or not you rent a booth, attending offline is very important. Ten online chats are not as good as one offline meeting. Trust is still the scarcest resource in this industry.
Bitcoin Orange Trader: L2s are receding, and PoW + AI is being discussed again
Bitcoin Orange Trader noticed at the venue that although L2s such as Blast and Abstract have shut down one after another, PoW projects have begun to be discussed again.
During TOKEN2049, one project hotly discussed in overseas circles was Pearl, which attempts to solve a long-standing problem: can AI computing power, while providing real inference services, also mine a Bitcoin-like currency along the way?
Bitcoin Orange Trader believes that PoW + AI has a chance to become a new narrative, but what determines how far PRL can go is not the technology itself, but whether AI companies are truly willing to pay for this computing power over the long term. Getting AI computing power to mine a coin is a technology, but making that coin an asset the market truly needs to hold is the real big business.
Joe Zhou: Stablecoins take center stage, and Crypto is moving from the virtual to the real
Foresight News reporter Joe Zhou has attended TOKEN2049 in Singapore for four consecutive years. His biggest impression this year is that stablecoins became the protagonist of this year's conference.
He observed that nearly one-third of the booths were related to stablecoins. Stablecoins have already formed a huge and complex industrial chain, and even small companies can become leaders in niche scenarios.
He heard many specific data points:
Some stablecoin payment companies have already enabled QR payments in 18 countries, with notable growth in Indonesia and Brazil;
A stablecoin wallet had more than 100,000 users completing payments in real scenarios last month, with an average monthly payment amount of more than 100 USD per person;
Stripe partnered with Shopify to let U.S. merchants accept stablecoin payments;
Stripe partnered with Meta to let creators in Latin America and Southeast Asia choose to receive platform income in stablecoins.
However, although AI Agent payments were discussed by many people, after talking with several stablecoin companies, he found that AI Agent payments still show no sign of large-scale implementation.
Another core observation from Joe Zhou is that Crypto has begun to move from the virtual to the real.
The most commonly heard phrase this year is "value capture." In the past, Crypto liked to talk about narratives, visions, TVL, and active addresses. But this year, people began to ask: Where are the users? Where does revenue come from? Who do profits flow to? What value can token holders actually capture?
He believes that five "won't-disappear" industries are taking shape in Crypto: public chains and DeFi, exchanges, stablecoins, Perp DEXs, and prediction markets. The common point of these industries is that users do not use them to support grand narratives, but because they provide clear financial services.
At the same time, however, the industry still has a habit of reinventing the wheel. There are more and more Perp DEXs, prediction markets, and stablecoin chains, and a new wave of chain-building is forming. Whether the market really needs so many public chains is still questionable.
His conclusion is: this year was the most restrained edition of TOKEN2049. Crypto practitioners began to stop telling stories and turned to talking about real business, real revenue, and value capture.
Crypto Chris: Beneath the surface prosperity, opportunities for ordinary people are shrinking
Crypto Chris attended TOKEN2049 for the first time. He did not go to the main venue and mainly attended side events and dinners. For him, the biggest takeaway was feeling once again that in the AI era, offline human connection will become more important.
Face-to-face communication is still the warmest way to interact and the easiest way to build trust. But he also felt a kind of "false prosperity." There were many attendees, even more than at Hong Kong Consensus and Web3 Festival, but the industry's own problems are also obvious.
Against the backdrop of a new cycle and accelerating integration with traditional finance, crypto-native narratives have almost disappeared. Everyone is frantically looking for opportunities, but opportunities for ordinary people are shrinking. Beneath the surface liveliness lies more anxiety.
He observed that the most discussed directions included: equities on-chain, Perps, prediction markets, and payments. These tracks are also merging with one another, for example Polymarket starting to do Perps and Hyperliquid starting to do prediction markets. On-chain options may become the next direction worth watching.
For KOLs, he believes the market has become harder: fewer projects, lower budgets, fewer ads, and it is increasingly difficult to rely purely on traffic and exposure.
What will be remembered in the future is still hard individual capability, including: a distinct personal IP, a stable business model, trading ability, content ability, and the ability to deepen one's work over the long term.
His personal strategy is to continue focusing on the primary market, do well in trading and content, while also improving his spoken English, exercising, and always going long on himself. In addition, he also mentioned that Singapore is very friendly to Chinese-speaking people and crypto people, and identity planning and overseas expansion will become long-term unavoidable topics.
Isabelle Payment: The truly useful information at conferences is often not on stage
Isabelle observed TOKEN2049 from a payments perspective. She originally thought there would not be many new projects this year and that attendance would not be too high, but because the Dubai edition was canceled and Singapore became the only flagship event of the year, combined with F1 weekend, the entire Marina Bay was packed.
The high-frequency words at the venue were still: RWA, Perp, AI Agent, and stablecoin payments.
But there were not many new things that truly stood out. In the past, everyone competed over who could discover the next opportunity first; now, more often, they compete for share within the same opportunity.
She believes the biggest change this year is that institutions have begun to "claim their own place." The appearance of traditional financial institutions such as Nasdaq, Franklin Templeton, BlackRock, and Morgan Stanley on the agenda shows that competition in hot tracks is shifting from discovering new narratives to fighting for real users, capital, and distribution channels.
Her most important observation is: the truly useful information is often not on stage, but off stage. On stage, people talk about global coverage, grand narratives, and trend judgments; off stage, what is discussed are the specific bottlenecks:
Stablecoin channels are stuck on local banking relationships and settlement cycles;
The order book depth of some Perps is fabricated, and it collapses as soon as it is withdrawn;
After AI companies try to collect payments in stablecoins, merchant reconciliation and refund processes cannot keep up;
The biggest problem for RWA projects is not the speed of putting assets on-chain, but whether accountability is possible after something goes wrong with the asset.
As a payments practitioner, she cares more about one thing: how exactly does money come in, how does it go out, and who is responsible when problems arise.
She also reminded that when project teams cooperate with market makers, they should ask clearly:
Are they using borrowed coins or their own funds?
How is the strike price of the options attached to borrowed coins determined?
Are assessments based on order book depth or trading volume?
Can trading volume be washed?
Can depth actually be sustained?These details won't appear in the main stage speeches, but they determine whether a project has actually built real liquidity or just put on a show.
She also specifically raised security concerns. On the opening day of TOKEN2049, trader Frogman reported that two wallets were compromised, with losses of approximately $4 million. Although there is currently no evidence linking this to on-site activities at the conference, it serves as a reminder to everyone: in a highly dense social setting, trust is built too quickly, and that itself is a risk.
Don't scan random QR codes, don't click unfamiliar links, don't casually connect wallets or sign authorizations. It's best to keep social devices and large-asset signing devices separate, and use a small wallet when testing new products.
Her conclusion is: what's truly worth mining at TOKEN2049 is not the next 100x coin, but the problems that haven't been solved yet.
Jade's RWA Notes: Without RWA, DeFi May Really Not Go Far
At TOKEN2049, Jade heard a line: Without RWA, DeFi is a dead end.
This line came from Superstate founder Jim Hiltner. Although it's a bit spicy, he believes this may be exactly the problem DeFi must face in its next phase.
Why might DeFi die? Because in the past, DeFi's TVL was almost a shadow of native token prices. TVL following coin prices shows that DeFi's scale doesn't depend on real demand, but on coin price fluctuations. The yield side has the same problem. In the past, on-chain yields mainly came from leverage demand for crypto assets. In bull markets, everyone borrows to add leverage, so interest rates are naturally high; in bear markets or when traditional risk-free rates are higher, capital leaves DeFi.
If 10-year U.S. Treasury yields are already attractive, safe, and large-scale, then on what basis can DeFi retain deposits? The answer is not simply offering higher yields, but introducing more high-quality real assets.
Jade observed that RWA has already begun to become an incremental source for DeFi. In Hyperliquid's HIP-3 market, perpetual contracts related to real-world assets have already accounted for a fairly high proportion of trading. On Solana, the scale of RWA assets and the number of holding addresses are also growing rapidly.
Asset types have also expanded from U.S. Treasuries and money market funds to private credit, reinsurance, royalty funds, and more. These assets have low correlation with the crypto market, precisely filling the gap DeFi lacks most.
But he also emphasized: RWA without DeFi is just an on-chain shell. After many RWAs go on-chain, they are not truly used. Among tokenized stocks and tokenized funds, the proportion that actually enters DeFi, becomes collateral, or participates in portfolio strategies is still very low. The RWAs truly suitable for entering DeFi are not necessarily all assets, but those that:
have relatively stable returns;
are not normally accessible to ordinary people;
can be used as collateral;
can generate sustained on-chain demand.
There are three main reasons RWA and DeFi have not truly connected yet:
Liquidity problem: RWAs often have T+1, T+2, or slower redemption cycles, making it difficult for on-chain lending markets to confidently accept them as collateral.
Interest rate structure problem: Institutions need fixed rates and fixed terms, but the depth of the on-chain fixed-rate market is still insufficient.
Institutional access problem: Wall Street will not enter DeFi as ordinary depositors, but as borrowers, issuers, and asset managers, and it needs KYC, permissioned systems, and legally enforceable arrangements.
His conclusion is: the phase of DeFi serving only crypto assets is coming to an end.
In the next phase of DeFi, what's being compared is not simple TVL, but:
which real assets can become collateral;
whether the fixed-rate market can achieve depth;
whether institutions can access it in ways they are familiar with;
whether RWAs are truly used after going on-chain.