Recently, I've been pondering a question: why is the cryptocurrency industry so pessimistic right now?

Projects are closing one after another, venture capitalists have shifted to hype AI, and even developers are flocking en masse toward AI. Fewer and fewer people are attending conferences, and the entire community is permeated with a sense of doomsday. "Still in crypto? Switch to AI now" has become the mainstream voice. But is this really the right choice?

Interestingly, if you look at the data, the crypto industry has actually never been more profitable. According to DeFiLlama's tracking, since 2018, native crypto protocols have generated $74.8 billion in fees. And these funds are increasingly concentrated recently—over the past 18 months (January 2024 to June 2025), they produced $31.4 billion, nearly half of the total.

This is the truly fascinating part. An industry that appears to be in decline is actually experiencing its best performance. What is hidden behind this contradiction?

I’ve analyzed the revenue flows. The astonishing discovery is that two stablecoin issuers—Tether and Circle—account for 34.3% of the entire crypto industry's fees. In other words, for every dollar earned by the industry, $0.34 flows to these two companies. Tether’s revenue is nearly three times that of Circle.

Why is this happening? First, demand. Countries in the Global South need to hedge against local inflation, and the US dollar (including digital dollars) fully meets this need. Second, cost structure. Blockchain bears all the costs; Tether and Circle don’t need to significantly increase staff for every additional $1 billion in issuance. The marginal cost of transferring funds on-chain is close to zero.

These two forces intertwine: demand drives issuance, and the cost curve flattens, ultimately forming one of the most capital-efficient businesses in financial history. Their moat comes from first-mover advantage—network effects generated by multiple exchanges connecting, which pure technology cannot replicate.

But stablecoins are just part of the story. The real growth driver comes from trading. Telegram trading bots, meme coin platforms, perpetual futures exchanges—these products contributed $575 million in fees by January 2025. Users are willing to pay high fees for quick profits. This category grew from 1% in 2022 to over 15% in 2025.

Applications like PumpFun, which aggregate underlying components and optimize user experience, have generated millions of dollars in revenue. They are not technically complex; their advantage lies in distribution and user experience. Hyperliquid dominates the decentralized perpetual futures market because its order book depth is comparable to centralized exchanges. Without this parity, users have no reason to migrate.

These "meme trading platforms" and perpetual exchanges are essentially packaging and selling risk. Some will evolve into core financial technologies, used worldwide for trading various assets.

But what about protocols? Those foundational layers that record all network fund flows? Their stories are entirely different. They are victims of a novelty premium.

In January 2023, Optimism’s price-to-fee ratio (PF) was 465x, Solana was 706x. Now, Solana has dropped to 138x, Arbitrum to 62x, OP to 37x. Polygon is closer to the valuation of a fintech company, at only 20x. These protocols now have more users, better liquidity, and more complex applications built on them, yet their valuations have declined.

Why? Because the market has realized something. Historically, Layer 1 and Layer 2 chains traded at extremely high premiums. If this premium had been well-invested, it could have created new economic systems. But open-source features and tokenization convenience have led us to have fifty copies of the same product across thirty networks. The result is a breakdown of composability, and the value of mechanisms like cross-chain bridges and messaging is declining.

Look at DeFi core projects to understand this. Investors have too many choices, lack innovation, and valuations have plummeted—even though these projects have indeed driven more economic activity. The market is highly fragmented; new projects like Kamino, Euler, Fluid have emerged, but their price-to-fee ratios are well below 2022 levels. Some exchanges now have ratios below 1, meaning their market valuation is less than the fees they generate in a year.

Here emerges a strange paradox: although the valuations of underlying protocols are falling, applications built on these protocols are generating higher revenues in shorter periods. Since early 2020, the number of teams earning over $13k quarterly has steadily grown, now exceeding 100. Protocols that took 24 months to reach $10 million annual revenue in 2020 now do so in just 6 months. Pump.Fun, from launch to $10 million monthly revenue, took only two months—setting a record.

But behind this lies a brutal reality. Today, nearly 900 crypto protocols generate revenue, each competing for an increasingly shrinking median share. The median monthly income has fallen to $13k.

Blockchain-native companies have three types of moats: first-mover advantage (Tether and Circle’s network effects are irreplicable), liquidity moat (Aave and Hyperliquid maintain deep liquidity across cycles), and distribution moat (relying on capital turnover and cyclical consumer demand).

But here’s the problem. When Visa’s price-to-sales ratio is 18x, shareholders receive dividends and buybacks, with legal ownership. When Aave’s P/S is 4x, token holders only have governance rights, with economic benefits only recently beginning. Hyperliquid’s buyback program brings HYPE holders close to equity owners—Aave has approved a $50 million annual buyback plan. But these are exceptions.

Most protocols lack mechanisms to return value to token holders. These low P/S multiples seem cheap, but the rights of shareholders are weaker than in traditional markets. This situation may arise because the crypto industry creates revenue at a scale and efficiency that traditional businesses cannot match. Small teams operating global financial infrastructure, with near-zero marginal costs and no physical offices.

Breaking down further, Aave’s P/S is about 4x, Hyperliquid about 7x—these are not bubble multiples, even lower than the closest traditional financial benchmarks. The only publicly listed large crypto exchange has a P/S of about 9x; CME Group around 16x; Visa about 15x.

Tether employs 125 staff and earns roughly $12.5 billion annually. This is the highest employee productivity in corporate history. Despite the overall 170x P/S seeming crazy, the market prices protocols that generate real income at levels equal to or below traditional financial infrastructure.

This raises a key question: what is the actual use of tokens? In crypto, entrenched duopolies have become the norm. Traditionally, founders needed to borrow or raise capital. But Hyperliquid, Uniswap, Jupiter have proven that with token incentives, people will invest capital.

Tokens may evolve into two functions: coordinating capital and resources from the right crowd, and granting governance rights. But tokens must have claims on economic activity and the ability to guide governance. Many L1 and L2 tokens struggle to fulfill both. Teams and VCs often hold most tokens, leaving retail investors in confusion.

Today, various protocols are trying to answer a long-standing question: why do people hold these assets?

For founders, this means rethinking what they build. Data shows all blockchain products ultimately profit from two core principles: either extracting small fees from high-frequency trading or taking large commissions on verifiable transactions. The advantage lies either in trading speed or in transparent verifiability.

The market will eventually move toward extreme efficiency. We see that 70% of market share in multiple segments is controlled by just two companies. This is the harsh reality of market operation. For founders, the funds that once flowed into their tokens are now redistributed into assets with higher volatility or higher capital returns.

Long-term capital does exist, and it may even pay premiums—if the value of the underlying business is recognized. Investors in Google and Amazon don’t rush to exit because their core businesses are inherently valuable.

In an era where even software value is questioned, crypto applications will have to find new ways to demonstrate value. Most long-tail applications like Web3 social, identity, and gaming are hard to scale and difficult to differentiate meaningfully from traditional competitors.

The era of building crypto infrastructure is over. It will merge with the internet. Soon, people won’t talk about "online" businesses—they will simply exist within the network. No one will call themselves an "mobile app developer"; you are the developer.

So, should we switch to AI? Maybe not necessarily. The key is to understand the real changes the crypto industry is experiencing. Projects that generate real revenue and have moats are actually valued reasonably. The problem is that most projects lack these qualities. If your ideas are based on real needs and sustainable business models, not hollow narratives, then the crypto space still remains worth holding onto.