Author: Jocy Lin, IOSG

This article is organized from Jocy's live keynote speech at Hong Kong Money Frontier 2026. For easier reading, we have made moderate edits while staying true to the original intent and corrected any verbal slips or abbreviations from the on-site speech.

1. Introduction: Know its white, guard its black

Hello everyone, I am happy to be invited by the Benmo community to gather with you all here in the bear market.

The topic I want to share this time is very interesting—Contrarian Investment, Frontier Betting: Why investments in a bear market often yield the greatest returns.

Let me briefly talk about the background. This should be the fourth bear market I have experienced. This time, many people's feelings are very clear, yet it is very different from the past. I really like a quote from The Tao Te Ching: “Know what is bright, guard what is dark”: knowing what is bright but willing to guard the dark to persevere. This in itself is a virtue.

Before getting on stage, many friends asked me: Is IOSG still investing?

We are still investing. The current feeling is very similar to 2018 and 2019—many people do not know how to invest. IOSG will continue to invest in the crypto market; we do not wish to become “the last gatekeeper of Asian crypto VC.” We hope to have more investors join Web3 to help more Asian entrepreneurs towards a better future.

Today, I will first clarify the big background using three “appetizers”—Bitcoin, stablecoins, Ethereum—then I will discuss four directions about real businesses that are making money from our frontline observations.

2. Bitcoin: Four-year cycle, overshadowing all narratives

Let’s start with Bitcoin.

The recovery from the lows of 2022 was driven by several catalysts exclusive to Bitcoin: the collapse of Silicon Valley Bank in 2022, the hype around spot ETFs pushing the price to $60,000, and Trump’s election pushing it to about $120,000. By the end of 2025, most of these catalysts will have already been fully priced in.

After that, the situation changed. As the AI cycle begins after 2025, assets related to AI, gold, and Nvidia are all strengthening; ideally, these conditions should be beneficial for Bitcoin, but Bitcoin is no longer responding to them and has dropped 29% from the beginning of the year until now.

Why is this the case? Borrowing a phrase from the Benmo community’s "Bitcoin divine power," it is very interesting: The four-year Bitcoin cycle once again overshadows all more complex narratives. The previous cycle framework clearly indicated that Bitcoin would peak in October 2025, and that’s exactly what happened; the subsequent bear market also began in October 2025.

Many people might argue that the four-year cycle could just be a coincidence. But we see that many crypto OGs reinforce this four-year cycle: as long as enough investors expect the market to peak at some stage of the cycle, they will collectively reduce their positions around the same time—this collective behavior, in turn, creates the cycle they expect. This also explains why, under favorable macro conditions, Bitcoin still turned down. The belief that “AI, gold, the index” can support Bitcoin is actually dwindling.

Therefore, I believe the cycle still holds. In the coming months, the focus should no longer be on explaining "why Bitcoin is underperforming," but rather on identifying whether the next round of accumulation phase has already begun.

Here, I also want to share some data research from IOSG. We compiled the duration of each cycle, and the results were remarkably consistent: since 2015, the upward duration has been around 1060 days. Based on this, we proposed several possible scenarios for this cycle:

• Scenario 1: Bitcoin could bottom out in the $45,000–$60,000 range around October this year (2026);

• Scenario 2: Slightly lower, falling to $40,000–$55,000 in Q1–Q2 of 2027;

• Scenario 3: Various other versions.

Predicting cycles can lead to many versions, but our internal data research points to the first version.

▲ Four-year cycle and bottoming scenarios

3. Stablecoins: A year as important as 1975 and 2001

The second “appetizer” is stablecoins. This year is a pivotal year for stablecoins, a very influential year.

I want to make two analogies.

On May 1, 1975, the U.S. Securities and Exchange Commission abolished the fixed commission system that Wall Street had implemented for 183 years, turning stock trading from a game for the wealthy into one where ordinary people could participate, ultimately giving rise to a $140 trillion asset management industry, with Fidelity and Vanguard being established afterward. In 2001, China joined the WTO, which was also the most significant event of change for the entire Chinese economy.

I believe that what we are witnessing today is the same event occurring again, only this time the stage is the stablecoin of the crypto industry. The GENIUS Act becoming law means that banks can now legally issue dollars on-chain. I believe that after the passage of this bill, crypto will reach a turning point, transforming from “high-risk assets” to “legitimate financial infrastructure.” Looking back in hindsight, 2026 will be remembered in history like 2001.

4. Ethereum: Moat, Challenges, and the Future Leadership Question

Next, let’s talk about Ethereum. This topic is complex; I have listed two tables: one for its advantages and one for its issues.

There are four advantages: as a settlement layer, it supports the largest stablecoins and RWA; safety and yield—most DeFi security and yield are on Ethereum; relying on Layer 2 scaling, the finality of the entire Rollup ecosystem settles back to Ethereum; and it has more developer infrastructure than any other L1.

However, there are also concrete issues: L2 has intercepted the fees originally paid to L1; the ETH/BTC price ratio has notably underperformed this round; and there are new public chains like Hyperliquid, as well as competition from off-chain.

But I think the core issue lies beyond these. Recently, I have written a lot about Ethereum on Twitter. My core view is that if we compare Vitalik with Elon Musk, many things Vitalik says often take 10 years for you to truly understand why he says them and why they can be realized. What we are most looking forward to is for Vitalik to be on the front lines like Elon.

Vitalik now also says he wants to turn Ethereum into a “smaller ship.” But frankly, from Tomasz's departure to the establishment of new institutions for Ethereum, all these ultimately point to a core question: Who will lead Ethereum? Who will govern? Who will manage? Which founder can stand on the front lines and make this decentralized, open organization great again? I think this is a very tough challenge.

As for stablecoins themselves, everyone is familiar with them: Ethereum primarily focuses on institutional funds and developed markets; Tron, on the other hand, primarily focuses on USDT, serving developing markets and cross-border dollar flows.

▲ Ethereum: Advantages and Shortcomings

▲ Distribution of Stablecoins by Chain

5. How we invest in the bear market: Redefining allocation ratios

Next, let’s talk about how IOSG executes investments in a bear market—this is also interesting because we have redefined the allocation ratios for investment during the bear market.

We have lowered the proportion of the primary market; while OTC, secondary market, and incubators have increased in proportion.

At the same time, we see a clear trend: Crypto is evolving from a trading market to an internet-native financial infrastructure.

• Layer 1, Better Money, Better Rails: Stablecoins have proven that better money and better settlement tracks can achieve internet-native implementation ahead of traditional systems.

• Layer 2, Internet Capital Markets: RWA and tokenization can provide clearer real returns and collateral for on-chain dollars.

• Layer 3, Large-scale Applications: I believe the crypto industry will run out the next ByteDance or Pinduoduo like the internet industry.

• Layer 4, AI Agent × Crypto: As financial infrastructure becomes internet-native, the ultimate users will no longer be just humans—agents also need wallets, payments, identity, and programmable ownership.

So ultimately, money will go on-chain, assets will go on-chain, and value will concentrate toward the application layer and interface layer, while AI agents will share the same infrastructure with humans.

6. Why there are the best investments in a bear market

Let’s return to the present. Why are there the best trades, the best investments in this bear market? Because in a bear market, quality is often mispriced, while real income can prove itself. Here are three key points:

First, valuation misalignment. At the worst emotional times, truly valuable projects may be driven down below their intrinsic value—the best entry points only appear when everyone else exits.

Second, real income can survive. We prefer businesses that can verify cash flow, not just narratives. If a project can weather a bear market—having clear customers and clear products—then it will amplify its growth in a bull market through compounding.

Third, withstand scrutiny. We only invest in projects that can clearly articulate their model from start to finish: who pays, why they pay, how much they pay. Income that can be externally verified is the real income.

In this round, we see real money-making projects in Crypto, with scales ranging from hundreds of millions to about $5.8 billion, distributed across DeFi and various infrastructures. Below, I will discuss four directions about the real situation we observe on the frontline.

Direction 1 · Stablecoins and Payments: Circle, fun.xyz, RedotPay

At IOSG, we are still holding four-hour IC (Investment Decision Meetings) every week. Some people say that in a bear market, what projects are worth discussing? In fact, there are many.

#Circle: Three sources of income, one DCF

Many people ask about Circle; let me explain its model. It primarily earns from three sources:

First, reserve income. About $73 billion in reserves is used to buy short-term debt, earning about $2.6 billion annually at an annualized rate of 3.5%, which will fluctuate with Federal Reserve policy.

Second, distribution costs. Circle shares its main income—about 62%—with partners like Binance and Coinbase, leading to a gross margin of only 38%.

Third, underlying long-term equity value. Circle has developed its own public chain, CCTP cross-chain, and developer API, which currently only accounts for about 6% of its income.

We have created a DCF valuation model for Circle: from stablecoin float → (yield) reserve income → (38% gross margin) stablecoin gross profit → (15 times in the third year, then discounted by 20%) per share present value.

#Fun.xyz: The Stripe checkout of the crypto world

Fun.xyz can be understood as the Stripe checkout of the crypto world—a universal deposit address (UDA), where all money going into Polymarket must pass through it; IOSG is also one of its investors.

#RedotPay: A crypto card that can be spent anywhere

RedotPay is currently the most popular crypto payment card on the market. A crypto card that can be used anywhere that accepts Visa and Mastercard, bringing crypto into real applications.

A few numbers: over 5 million cards issued, globally usable, with TPV of about $3 billion, leading the market by 4 times, with annual revenue of about $150 million. So our underlying judgment is that products like RedotPay, which combine payments and stablecoins, are moving towards the mass market and have already gained recognition.

▲ RedotPay Scale and Capital

Direction 2 · Prediction Markets: Niche futures are mainstream futures

The second direction is prediction markets, which are moving towards the mainstream. Niche products will migrate to mainstream products, the niche future is the mainstream future, and this is the cold start phase that helps crypto towards large-scale applications.

Let’s look at the scale of two industry leaders: Polymarket had a trading volume of $26.2 billion in Q1 2026, up 90% quarter-over-quarter; during the World Cup (from June 11 to July 19), the trading volume exceeded $15 billion. Its differentiation lies in globalization and non-custodial, representing one of the best next-generation crypto applications. Kalshi is taking a compliance route, having obtained a federal license; its trading volume in Q1 2026 is about $32.1 billion, and it distributes through Robinhood and Interactive Brokers. It was established in 2018 and has faced doubts for years, but it survived the bear market, obtained its license, and has shot the steepest revenue curve we've seen in American fintech.

▲ Polymarket · On-chain route

▲ Kalshi · Compliance route

Direction 3 · AI × Crypto: How computing power, data, and money flow

The third direction is AI and crypto.

In the past five years, everyone has focused on model quality: whose transformer is better, whose RLHF is smarter. But now we need to think about a more core logic: where does computing power come from? Where does data come from? How does AI's money flow?

Let’s first talk about computing power. Crypto has proven one thing: open networks can coordinate hardware resources globally. During Ethereum's PoW era (before The Merge in 2022), the GPU computing power gathered by the entire network was on par with a cutting-edge training cluster. This does not mean that miners' GPUs can be directly used to train cutting-edge models—different calibers—but it proves that: with token incentives, you can aggregate globally dispersed idle hardware. If this mechanism is correctly applied to the AI computing power market, it presents a real opportunity.

Secondly, pertaining to data, and thirdly, how AI's money flows, which refers to the model of agent banking. Here are a few specific cases.

Grass has about 8.5 million users who earn points by sharing unused bandwidth through a browser plugin and App. Its network layer distributes scraping tasks from AI labs to these nodes and then cleans and structures the webpages into enterprise-level data. The key architecture is that it can distribute the demands of AI demand side (e.g., companies like OpenAI and Anthropic that are willing to pay for training data) to on-chain users, incentivizing them through tokens or revenue. Currently, its data volume exceeds 250 PB. The economic model is also quite real: projected revenue of about $17 million in 2025 and expected to exceed $70 million in 2026. So you see, crypto projects that issue tokens have become very real, generating income, cash flow, and buyers willing to pay for them. First to B, then to C is also one of our investment philosophies.

Hyperbolic is another IOSG project working in the inference and GPU computing power market. Over 250,000 developers are building on the platform, with clients including some cutting-edge AI labs.

The third typical project turning Crypto into AI is Nous Research, which you may know as Hermes (“lobster”). I won’t delve into its entire tech stack due to time constraints.

▲ Grass Economic Model and Data Flywheel

▲ Hyperbolic Computing Power Market

Direction 4 · On-chain Trading and Credit: Collector Crypt, Hyperliquid

The fourth direction is on-chain trading and on-chain credit, which is quite interesting.

Collector Crypt is akin to a “pawnshop + card store,” only it operates on-chain. It is now the second top application on Solana by revenue. The model is very simple, consisting of three steps: the first step is the supply side, purchasing and custodizing real cards and storing them in a vault; the second step is on-chain tokenization, turning each card into a tradable token that can be redeemed anytime; the third step is to liquidate and exit. Currently, its cumulative trading volume exceeds $1 billion, with monthly active users at 4 million; its daily revenue is among the largest applications on Solana; protocol revenues were approximately $7.2 million in April, about $9 million in May, about $15 million in June, and around $12 million in July. At this scale, it has a chance to exceed $200 million in revenue this year. This is a project with real income, has issued tokens, and has clear demand.

Hyperliquid is also very interesting; many people say it is “eating Binance.” To put it simply, it is a self-compounding buyback flywheel that can be understood as a decentralized Binance doing futures: each transaction incurs a fee, of which about 97% goes to buy back tokens; more transactions lead to more fees, and more fees lead to more token buybacks, and so on.

▲ Collector Crypt · On-chain cards

▲ Hyperliquid · Buyback flywheel

7. IOSG's Confidence: Bridging East and West + Research Moat

Why us? Because IOSG has the best reach for global opportunities—most of our portfolio spans both East and West, North America and Asia; we can also capture the core paradigms of each cycle very early. Supporting all this is our very solid research—deep research is our investment moat.

8. Conclusion: The bear market is the starting point for layout

I have taken up a lot of your time; let me conclude.

The best investment opportunities are often hidden within the worst emotions. To emphasize: according to “Bitcoin divine power,” Bitcoin may reach a low point around the end of October this year. Therefore, those who have turned to AI or US stocks can also revisit the new opportunities in the crypto market.

So what will we seize?

First, projects with real income. Directions related to stablecoins, payments, AI and crypto are generating new, verifiable cash flows.

Second, businesses that have been wrongly killed this winter. Many projects have excellent fundamentals but are priced as if they have completed about 90% of a bear market—this pricing misalignment is an opportunity in itself.

Third, IOSG has captured them early. The goal of three rounds of contrarian and research-driven investments is just one: to identify the winners before the bull market arrives.

The bear market is the starting point for layout. Thank you all.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。