10K Ventures Annual Report (Part 2): Outlook 2025

原文作者： 10K 风险投资

This article shares our thoughts on 2025 and does not constitute investment advice.

It has been exactly half a year since the 10K Beta Fund closed its fundraising in May. Thanks to the support of all LPs, we were able to raise funds quickly in the first half of 2024, allowing the 10K Family Friends Fund to run fast.

2024 is a year full of turning points and hope for us. We and our portfolios have taken this opportunity to make remarkable progress in the past six months and gradually expand our ecological layout. 2025 is believed to be an even more promising bull market year. In the previous article, we reviewed the industry trends in 2024. This article will share our predictions for the industry in 2025, mainly including:

(1) Compliance

(2) 交换 structure

(3) The future of AI agents

(4) Changes in cross-border payments brought about by stablecoins

(5) BTCFi’s leading competition

(6)Why Application not Infra

01. Compliance becomes even more important

After FTX, the industry has become less and less tolerant of opacity. For example, in the field of exchange custodial wallets, trading venues/fund custody have become more and more like the traditional separation of powers and responsibilities model. For example, many institutional clients money is now stored in ceffu, and transaction matching occurs in Binance. Exchanges all disclose their proof of reserve.

At the beginning of 2024, in the United States, since the BTC/ETH ETF was gradually passed, we have seen that the Trump regime has become more friendly to 加密货币, and Gary Genler is about to step down in January. In Asia, HK started the stablecoin sandbox in April this year, giving companies such as RD/JD.com the opportunity to initially explore stablecoins. In addition, OKX became the worlds first exchange to obtain a full operating license in the UAE in 2024.

Let’s not discuss some very abstract macro policies for now, but if some changes can be made to the token utility during Trump’s term, we will consider this to be a great benefit to the industry’s copycat projects.

Currently, the utilities accepted by the SEC are: 1. Gas fee; 2. Staking (but LST is Security); 3. Governance

If the SEC can clearly accept tokenomics in the future: 1. Revenue sharing; 2. Buy back; 3. ETH/SOL ETF can be staked/restaked in the future. Once the SEC is able to make changes, Asian crypto centers such as HK/SG will continue to follow up on the policy, which is a great benefit for copycat projects.

2. Changes in CEX spot market

I have always said that CEX, as a public traffic market and the flywheel effect of Maker + Taker, makes it difficult to change the competitive landscape once it is formed. There are only very few events that CEX can defeat Binances flywheel effect, such as all the middle and senior executives of BN being wanted by the FBI, and Binance being exposed for misappropriating user assets to go long and short in the market.

But after 2-3 years of evolution, we think:

2.1 The penetration rate of the Korean market can at least double to 60%

Based on the statistics of the number of investors who have accounts on the five major Korean exchanges, including Upbit, Bithumb, Coin One, Cobit and Gopax, the Bank of Korea concluded that as of November this year, the number of people holding cryptocurrencies on Korean exchanges exceeded 15.59 million, an increase of 610,000 from 14.98 million in October. Currently, the total population of South Korea is estimated to be around 51.23 million. This means that the number of people holding cryptocurrencies is equivalent to more than 30% of the total population.

There is an even more interesting data. According to statistics, in August last year, the number of active stock accounts in South Korea exceeded 50 million for the first time. By February of this year, the number of active stock accounts in South Korea had exceeded 60 million, 8 million more than the total population, equivalent to 1.16 times the total population.

Traditional large institutions and whales in South Korea have not yet fully entered the market. As a market with XXX/KRW as the trading pair, we believe that the penetration rate of the Korean market will increase significantly next year under the extremely crazy bull market and wealth effect. The most typical case is the volatility and trading volume of UXLINK in South Korea around Christmas, which is so exaggerated that it is outrageous. It is strongly recommended that the CEO of the project party speak Korean in addition to Chinese and English and go to South Korea for on-site CX.

2.2 Bitget, the Pinduoduo in the cryptocurrency world, is expected to become one of the top three CEXs

BGs trading volume this year has grown from 500-1 billion USD per day at the beginning of the year to several billion USD per day in December (only about 20% of the trading volume is contributed by the great stand-alone coin BGB, the rest is fully competitive). The market share has increased from 2.8% at the beginning of the year to 7.8% in December, becoming the worlds fourth largest exchange for non-USD trading pairs, surpassing OKX.

What Bitget did right, or its future growth comes from:

1. The core management team represented by handsome boss + Grace has a clear idea. Based on the empowerment of platform coins (for example, 1-2 launchpool + poolx per month), Bitget is continuously empowered around investment (FV) + media (FN) + Bitget Wallet (wallet) + Morph (L2). It makes me see the shadow of Binance in 20-21. The projects invested by FV are constantly listed on BG; now the token BWB of Bitget Wallet has been replaced by BGB; is it possible for Morph, a consumer-grade L2, to break through in the future?

2. The employees are really competitive, working hard every day, and the pay is really high. They poached people from Bn, OKX, and bybit with 30%-50% higher salary.

From the perspective of the market share of on-chain DEX, thanks to the emergence of Pumpfun, Raydiums market share has greatly increased, once occupying 28% of the entire chain DEX market share.

2.3 Bybit, the invisible winner benefiting from VC coins

Bybits market share was 8.79% at the beginning of the year, 16% in the middle of the year, and 11% at the end of the year, making it the third largest exchange after BN and Upbit. This year, many VC projects could not go to Binance, so they found that Bybit could also have a large trading volume. After going to Bybit, it is also a good choice to go to Korea Exchange + Binance. In addition, we have recently seen that Mantle has begun to take action to recruit Kols to cooperate with publicity.

After 25 years, can Bybit continue to move to the next level and become a small Binance?

3. AI solves the core contradictions of the industry

At the end of this year, during a conversation with a senior in Shanghai, a question was raised that made me ponder: what core contradiction has the industry solved?

I realized that figuring out this question would also give our fund an answer to the industry: how should we face the next primary market? I firmly believe that In trustless we trust . If the essence of the AI era is that people have acquired low-cost advanced intelligence, then Crypto will likely help open source AI. Mathematical trust achieved through cryptographic principles and code can solve core problems such as: (1) How to trust agents (2) How agents interact with each other (3) How the agentic economy unfolds.

We believe that great opportunities will arise in AI+Crypto. This is because the essence of encrypted networks is to build and transmit trust networks. Encrypted networks not only provide agents with a reliable trust foundation through cryptography and code, but also their decentralized open nature and inclusiveness of subject identity 定义nitions create a digital space for agents to act autonomously, fundamentally solving the lack of identity and trust dilemma of agents, and opening a door to true autonomy for agents.

3.1 Agent Economy

We believe that Agent Economy will be the most exciting direction in the AI+Crypto track. AI Agent is reshaping the boundaries between technology and business. This is not just a technical concept, but also a new form of productivity and organizational method. Multi-agent collaboration is a key feature of future technology architecture. Imagine an ecosystem composed of hundreds of professional agents, such as:

• Dedicated Solidity Programming Agent

• Agents that are good at Python development

• Project management agent responsible for overall coordination

These agents may not know each other directly, but through the central coordinator, they can achieve efficient coordination of complex tasks. This organizational form is similar to an intelligent container orchestration system and will completely change the traditional working mode.

The boundaries of business imagination are constantly being expanded. Taking IP operation as an example, AI Agent can achieve unprecedented innovation models:

• Acquire IP with historical value at low cost

• AI Agent acts as an IP broker

• Precision marketing on social networks

• Explore diversified monetization paths

3.2 On-chain autonomous agent, intelligent system that thinks and executes autonomously

In the Web3 field, the potential of AI Agents is equally exciting. For example, the Crypto Enabled AI Autonomous Agent has a wide range of tasks. It may be a simple USDC payment, or it may require the autonomous deployment of a batch transfer contract to save gas fees, or it may involve multi-step on-chain interactions and complex contract creation. Although many operations can be implemented with technical indicators, we cannot foresee and enumerate all scenarios. Therefore, a truly autonomous agent needs to be able to understand the task, decide on its own which tools to use, what information to obtain, and finally convert it into a multi-step action plan, execute operations, and finally complete the task.

Autonomous agents that autonomously perform tasks in the field of encryption driven by AI models. The ideal of the technology is to build intelligent systems that can think and execute autonomously. An ideal AI Agent should have:

• Deep understanding of complex tasks

• Autonomous task decomposition

• Write implementation code independently

• Ability to integrate interdisciplinary knowledge

• Autonomously complete smart contract deployment

• Conduct complex on-chain asset management

• Realize cross-chain intelligent operation

3.3 AI agent + payment, the economic model of human-machine collaboration in the future

As a large and complex Agent ecological network is formed, the interactions between Agents and between Agents and humans will become more frequent, in-depth and diverse. Payment is one of the foundations of the Agent economic network, but the existing technology stack and payment system cannot support Agents to make autonomous payments. Agents need cryptocurrency payments.

In a decentralized encrypted network, these obstacles faced by agents are almost solved. The permissionless network and open ecosystem greatly expand the capabilities of agents. Therefore, I think that among all the directions of integrating AI and encryption technology, enabling AI agents to use encrypted networks is a direction that is very close to real needs and has practical prospects.

In the short term, giving Agents the ability to make more convenient crypto payments is enough to expand their scope of action. In the medium and long term, digital identities based on crypto networks, permissionless protocols, decentralized applications, and the capabilities granted by smart contracts will enable Agents to operate autonomously in the digital world, cooperating, competing, and gaming with humans and other Agents. Crypto networks not only provide Agents with the basis for autonomous action, but also open up new possibilities for the future economic model of human-machine collaboration – when Agents can participate in economic activities autonomously, this development will bring about tremendous changes to society. (The above is written by our LP Wang Chaoge, it is so well written that I have to put the original text up.)

3.4 AI Data, do the left-wing and right-wing approaches lead to the same goal?

With our first round of heavy investment in Sahara in 2023 and Portfolio Vana listing in BN/Upbit in mid-2024, we are more aware of the importance of AI data to AI companies. Nowadays, algorithms and computing power are no longer the core factors that restrict model capabilities. From 2023 to the beginning of 2024, computing power was very tight, and AI companies had a very high demand for A 100/H100, and it was difficult to find a machine in China. A large number of A 100/H100 were smuggled into the country, and cloud vendors rented machines for 1-2 years. Therefore, a large number of decentralized computing power projects were born in 2023, such as IO/Aethir.

With the gradual evolution, the market has begun to talk about decentralized algorithms and token incentive developers to contribute model algorithms for revenue sharing. We will not comment on the reliability or story of this batch of projects. You can think about whether it is reliable. On December 27, 2024, Deepsake, a subsidiary of Huanfang, was born. It only used 1/10 of the pre-training cost of other large models in North America, and trained the ability of various indicators that are not inferior to the top closed-source large models in North America, with stronger reasoning ability and lower cost. This also gradually shows that algorithms are no longer the core factor restricting model capabilities.

We always believe that the only thing that can really restrict the models capabilities is data. We believe that in 2024, both web2 and 网络3 AI data companies will highlight their value.

In the world of web2 AI, there are more and more leading players in the data track. Scale.AI has become the leader in the data annotation track. Synthetic data companies represented by Lightwheel Intelligence have made efforts in the fields of embodied intelligence and autonomous driving, and have won many orders from large overseas companies.

In the world of web3 AI, our Portfolio Sahara is also the same. In 2024, it won orders from domestic top model companies Zhipu/Dark Side of the Moon/Ess Technology, as well as overseas top AI companies such as snapchat and Microsoft. At the same time, it also won orders from a certain department of a sovereign state, and progressed rapidly. Sahara is the most right-wing link in the AI Data track. It is extremely top-down. After receiving the order, it distributes it to users for annotation. The current alpha test is more about long text annotation. For details, see the Sahara screenshot in 1.3.3. Vana is the most left-wing link in the AI Data track. VANAs data comes from user contributions in the ecosystem. VANA ecosystem participants contribute social media or IoT data such as X and LinkedIn to DataDAO, and these data will be securely stored off-chain. After verification, cleaning and marking, the data is used for model development.

Whether companies in the AI data track can transcend cycles and become evergreen companies like Chainlink still depends on whether the company has the long-term ability to commercialize (2B to get orders) and listen to the community (2C, making products that users can perceive), rather than the ability to tell stories.

4. Stablecoin is changing cross-border payments

Fiat-collateralized stablecoins are changing the payment industry from top to bottom. Stablecoins have a dimensionality reduction impact on core issues such as the settlement cycle and transaction friction of traditional payment channels.

在 the end of this year, we are working on a fiat-collateralized stablecoin company. It is the team that I have seen in the past two years that is most likely to make a name for itself in the field of web2 cross-border payments. In the crypto-native stablecoin track, I have talked about Ethena and Usual in the past two years. I think ethena may have systemic risks and its token utility is too weak; Usual is a decentralized Tether, which is a bit out of place. Its returns are not as good as those of pure delta-neutral Ethena, and its AUM is not as good as FDUSD. In the fiat-back stablecoin track, we talked in depth about FDUSD and RD. We think each has its own problems. The problem with FD is that the team is neither crypto-native nor understands traditional payments. The problem with RD is that it is not crypto-native at all.

But until recently, after we talked with the executives of this team, I saw the real use case of crypto native + web2 cross-border payment that met my psychological expectations. After my interviews with the teams executives and upstream and downstream (payment companies, bulk traders), lets summarize and share some interesting points:

1. Issues that need to be addressed in the traditional trade industry

In traditional trade, the settlement cycle is a headache. It is necessary to arrange the flow of funds, determine the arrival time, the shipping settlement time, and the next batch of collection and payment after the leaflet arrives. The remittance time is usually 1-5 days, and the friction is 1%-3% due to the strength of the legal currency. When importing or exporting to some countries with small currencies, you need to pay attention to the real-time exchange rate. If it is a country with serious conflicts, it is quite risky for both parties. In the past, transactions have always been made in US dollars, but there are problems with cross-border transactions in US dollars, and there is also a need to worry about time differences. In addition, the intermediary bank is your own bank and the other partys bank. Sometimes there is insufficient liquidity, and payment and collection are usually delayed, making it difficult to calculate the cost of capital flow. There are also exchange rate deviations, and a series of problems where the goods have arrived but the payment has not yet been received.

For example, one of the suppliers of an international trader we surveyed in Africa had a terrible local financial system. They couldn’t even trust the banks and only trusted USDC because it could be exchanged for U.S. dollars through CB.

Another trader we surveyed had a client in Latin America tell us that after their clients paid U.S. dollars into their Argentine accounts, it was difficult to transfer the money out because Argentina was in great shortage of U.S. dollars. It was easy to transfer dollars into the banking system, but the process for remitting dollars out was very complicated.

2. The demand for stablecoins is completely bottom-up, and bankrupt payment institutions accept stablecoins

Initially, there were not many demands from customers. With the development in the past two years, the demand has gradually increased. When I went to Yiwu for investigation, I found that there were many small vendors in Yiwu. They would give foreigners two QR codes, one for payment and the other for the stablecoin address. You will find that its use scenarios are increasing.

Especially in some emerging markets, stablecoins are actually products that replace the US dollar. Many local users will accept stablecoins, which are not of the same order of magnitude as the US dollar in terms of bank accounts or liquidity.

Therefore, after many users at the end have U, for Chinese companies going overseas, when they go overseas in Africa or the Belt and Road Initiative, local suppliers will give them money to let them choose local legal currency or stable currency. I believe that many companies will choose to accept stable currency. This is a bottom-up process. As the popularity of the following becomes more and more widespread, these companies are gradually accepting stable currency. And in the process of collecting payments, he found that stable currency can indeed solve the payment problem, so he tried to force the payment institution to ask them if they are willing to accept this thing, and even help them deal with matters such as foreign exchange settlement. Finally, the payment institution began to reflect and push back, and began to handle related business.

3. The competitiveness of different fiat-collateralized stablecoins in cross-border payments in web2 comes from localized 2B and whether they can serve customers in real time

Being able to find a team when something goes wrong is a core competitiveness. Most traditional customers and traders know nothing about U, and their concept of U is another dollar. Therefore, this process involves whether the product can make people who have never used it understand how to operate it immediately, whether the operation is safe, and the service providers maintenance of the customer relationship with the customer, which are all very important.

What customers are most worried about is that they pay the money but don’t get a response for half a day, and they really need to pay the goods today. At this time, you tell me that the cash flow is not enough, and we need to delay for another day for this or that problem. Delaying for another day will cost an extra fine of 20,000 or 30,000 yuan.

5. Who will become the leader of BTCFi?

Continuing with what we said in our November quarterly report, we have always attached great importance to how to provide risk-free interest to Bitcoin holders in the Bitcoin ecosystem (and congratulations to our Portfolio Solv on Binance). Based on the first principles, we believe that this is the most essential need of Bitcoin holders. Based on this, the first direction we are bullish about is Bitcoin (re)staking interest, and the other is how to cross Bitcoin trustless to other chains to provide liquidity (the result of cross-chain is also to provide liquidity for financial management).

We will not go into too much analysis on the fundamentals of Babylon and Solv, but we would like to make a few points:

1. Solvs income comes more from the real trading side, such as funding rate arbitrage, JLPs Delta neutral and BTC coin-based options arbitrage. This type of income is very high in a bull market, starting at 15%-20%, but the yield in a bear market will be relatively low, starting at 3%-5%.

Babylons revenue mainly comes from customers own coins and BBNs coins. BBN may have a flywheel effect in a bull market, but in a bear market, Babylon ecosystem may have very poor revenue.

2. During the bear market, we need to look for financial income in traditional web2 scenarios. Currently, we are looking at scenarios such as supply chain finance, which may provide opportunities.

3. The first-mover advantage will be significant. As the leaders in the field, whoever can issue coins first and bring about a considerable wealth effect (marketing) will be the first to occupy the minds of users.

4. From the growth of ETF holdings, it can be seen that more and more BTC are in the hands of large institutions. Babylon/Solv, whoever can be more compliant and better meet the needs of large institutions, is more likely to become the first BTCFi in the long run.

6. Application ☑️ Infra ❌

Speaking of this topic, lets take a look at how much it costs to build a good ecosystem. Hyperliquid has grown from 3 yuan to 30 yuan, and the conservative estimate is that it cost $100 M+. In this cycle, the funds required to build a public chain top-level ecosystem have far exceeded the cost that ordinary infra projects can afford. We do not think that the underlying blockchain technology has a high possibility of paradigm breakthroughs, and the winning rate of early VC betting on infra projects is indeed too low to bear. We will be cautious about early investment opportunities in the infra field and will bet on applications.

But looking back at the performance of Application in 2024, it is really impressive. This year, the record of days that revenue break $100m has been broken continuously, first by Ethena (8 months), then by Pump.fun (7 months), which shows the extremely strong explosive power of the blockchain industry – as long as the direction and product are right, the market will give you the fastest response. But have you ever thought about why the following lists are all related to transactions?

Because of the right to keep accounts, the blockchain industry is all about the right to keep accounts. After the first barter transaction tens of thousands of years ago, transactions and bookkeeping have been passed down along with human civilization, from single-entry bookkeeping to double-entry bookkeeping during the Renaissance, and now blockchain bookkeeping, which are essentially more decentralized and distributed to everyone.

(The following are the original words of our LP, which are so well written)

The most mainstream assets are generated from the most important power. Bitcoin is used to account for the scarcity of fixed output, Ethereum and TRON are used to account for currency codes, which are actually mainly US dollar stablecoins and related transactions, and Sol is used to account for high-frequency trading memes. So the question is, this industry is currently only accounting for about 0.3% of the worlds assets and less than 1% of transactions, and the remaining 99.7% of assets and 99% of transactions have not been accounted for. This is the real beta and alpha. RWA should think from this perspective. Whats more interesting is that the so-called tokenization and valuation that cannot be analyzed by traditional fundamentals are to give a premium to this accounting power. This premium is the only means to ensure that this power will no longer flow back to the traditional system, that is, the inherent greed of human nature.

Looking back at the applications that have broken the circle in recent years, they are actually the accounting upgrades of assets in different application scenarios – DeFi (accounting of financial assets), GameFi (accounting of game assets), NFT (accounting of artworks/pictures), Friend.tech (accounting of quantified social value). Meme (accounting of spinach). As the underlying technology of blockchain continues to iterate and a large number of commercial capabilities and products are verified in 20-24, we may see the accounting upgrades of new assets in 2025 and beyond, such as AI x Crypto (accounting of computing power and data), DeSci (accounting of science/charity), RWA (accounting of houses/offices), PayFi (accounting of supply chain finance), and SocialFi (accounting after quantification of social value).

7. Traditional giants begin to enter

This year, we have observed that more and more traditional companies are entering the crypto market from various angles. For example, Paypal has entered the stablecoin market through PYUSD; JD.coms stablecoin has entered the Hong Kong stablecoin sandbox; a large number of US listed companies have begun to reserve BTC; traditional cross-border payment companies have begun to accept stablecoin payments.

Does this bring us a new round of thinking: How to serve these new players entering crypto? Will these new players have new ideas? Our focus next year will be to observe the demand for crypto from these large-scale new players.

8. What do you think?

In the process of preparing this annual report, we collected opinions from some friends in the industry on primary and secondary investment opportunities for next year.

The following are their main points for your reference:

1. The combination of AI and Crypto is still the direction that most people think has potential in the future, especially AI Agent, which is regarded as an important application in the next bull market. Multiple AI Agent coordination, the shift of Web2 AI startups, the combination of AI and MEME, AI and payment, AI and TEE, AI and stablecoins, etc. are all exciting.

2. Payment and RWA

3. New public chain ecosystem, focusing on Hyperliquid, Monad, Berachain, Base and Solana

4. ICOs are back

5. BTC is a deterministic opportunity

6. Risk-free arbitrage

7. DEX trading terminals (such as Dexx, GMGN, Infinex) and intent-based chain abstractions (such as Across, Near Intent for Base and SOL)

8. Options related, can have real income and arbitrage inside and outside the circle

9. Combining user acquisition tools with MEME, MEMEFi

This article is sourced from the internet: 10K Ventures Annual Report (Part 2): Outlook 2025

Related: Grayscale Report: From Virtual Influencers to Autonomous Payments, the Impact of AI Agents on Crypto Popularization

Original author: Michael Zhao Will Ogden Moore Original translation: TechFlow In the future, AI agents are expected to revolutionize the way we interact with the world. They will be able to perform a range of tasks for us that have never been possible before. However, to truly unleash the potential of these digital entities, they need not only strong intelligence but also economic autonomy. Fortunately, blockchain technology provides an ideal solution for this – as demonstrated in a recent experiment with AI influencers. The so-called AI influencers are chatbots that run autonomously on social media. These robots can not only manage their own blockchain wallets, but more importantly, they can understand economic incentives and use resources rationally to achieve their goals. Grayscale Research said that as AI increasingly uses blockchain…