Author: Christy Choi

Translation: Shenchao TechFlow

Introduction by Shenchao: Christy Choi served as a core executive in the early days of Binance Labs and has been deeply involved in the crypto industry for ten years. She now manages a fund that spans Asia, the Middle East, and the United States.

In this article, she presents a judgment: the crypto industry is undergoing a fundamental transformation driven by two simultaneously emerging structural forces—institutional capital entering through stablecoins and AI driving the cost of building everything close to zero.

She believes that the era of token speculation is coming to an end and that the winners of the next era will be infrastructure companies that hold licenses, have real revenue, and can serve AI agents. Her viewpoint is clear and suitable for comparison with the current state of the market.

The full text is as follows:

Some things have changed in the past twelve months, but most people in the market have not yet realized it.

I have been in the crypto industry for ten years—serving as a core executive in the early days of Binance Labs, making early investments and project developments through multiple cycles, and now managing a fund with a focus on Asia, the Middle East, and the United States. I have experienced every version of this industry: the ICO frenzy, the DeFi summer, the NFT bubble, and the cascading market crashes. Each cycle felt different at the time, but the same engine was driving it—speculative capital chasing narrative tokens.

This engine is dying. Not because crypto has failed, but because it has succeeded. What will replace it will fundamentally reshape this industry more than anything since Bitcoin.

Two structural forces are converging simultaneously: institutional capital entering through stablecoins and AI driving the cost of everything in crypto close to zero. Together, they change not only which tokens will win but also what crypto itself is.

Institutional Turn: Stablecoins Devouring Everything

Most crypto-native players have yet to internalize this: the largest wave of capital ever to enter this industry will not buy tokens; it will use stablecoins.

Stablecoins now settle trillions of dollars annually. They are the first crypto product that institutions, corporations, and governments genuinely want—not for speculation, but as infrastructure. When a multinational company manages its funds via a stablecoin framework, when a remittance channel switches from SWIFT to USDC, when a new bank offers stablecoin-denominated savings accounts to the unbanked in Southeast Asia—these are real economic activities moving onto the chain. Not TVL mining, not governance token speculation, but revenue.

This changes the entire value chain. The winners in the new landscape are not protocols with clever tokenomics but licensed companies with regulatory moats. Issuers of stablecoins, compliant middleware providers, licensed new banks, and settlement infrastructure—these companies are capturing the largest share of the institutional wave. By Crypto Twitter's standards, they may seem boring, but they will generate the most lasting returns of the next decade.

The moats of these enterprises are not technology; they are regulation.

This is where crypto-native players have always underestimated. In the old crypto world, moats came from liquidity, network effects, and community. In the new crypto world, the deepest moats come from holding a license. Every jurisdiction that finalizes stablecoin rules, tokenization frameworks, or digital asset banking regulations creates a window—typically 12 to 18 months—during which the advantages established by the first licensed operators are impossible for later entrants to replicate, regardless of their investment. Customer relationships, bank partnerships, compliant infrastructure, regulatory trust—these cannot be forked, nor can your agent write them out. By the time competitors obtain licenses in the same jurisdiction, the early movers have already locked in distribution channels.

This is happening on a jurisdiction-by-jurisdiction basis, not globally uniform. Europe's MiCA, the emerging stablecoin frameworks in Singapore and the UAE, South Korea's digital asset framework, and the regulatory framework now forming in Washington—each is creating different regulatory moats in different markets. The companies that will win are those that treat regulation as a strategic asset rather than an obstacle. They are hiring regulatory officials, not just engineers. They are shaping frameworks, not just compliance. Policy proximity—the ability to influence rules during the rule-making process—is the most valuable and least understood competitive advantage in the crypto industry today.

The native token speculation playbook—issuing tokens, attracting TVL, leveraging narratives to pump prices, unlocking and cashing out—is coming to an end because the capital entering the system does not play this game. Institutional allocators want yield, compliance, and predictable revenue. They do not want governance rights over a protocol that might be forked next quarter.

As real cash flow moves onto the chain, tokens inevitably begin to resemble equity. When a protocol generates real revenue and distributes fees to token holders, tokens cease to be speculative tools and become machine-readable proof of ownership of real businesses. This is convergence. It is not that tokens replace equity or that equity replaces tokens, but that both collapse into the same thing: programmable, composable, instant settlement rights claims to real economic activities. The shell no longer matters; what matters is that the underlying business generates cash, and the claim to it is machine-readable.

The same license moats apply to tokenization platforms. As stocks, bonds, and structured products move onto the chain, the platforms that tokenize them will not be permissionless protocols but licensed securities intermediaries operating under specific regulatory frameworks in specific jurisdictions. The infrastructure is crypto-native, the business model is traditional finance-grade, and the moat is still the license, not the code.

Crypto has spent ten years building the track for transferring value. But the track for transferring identity, qualifications, and authorizations is yet to be built. This is the gap—and where the next wave of infrastructure will be constructed. Teams that resolve authentication of identities at the layer of on-chain primitives, machine-verifiable compliance, and portable professional qualifications are building the layer that connects institutional finance and autonomous AI. Both sides cannot operate well without it.

AI Turn: Building Becomes Cheap, Validation Becomes Precious

The second force is AI, which has a far deeper impact on crypto than the "AI × Crypto" narrative suggests.

First, the obvious: AI has reduced the cost of building any software to extremely low levels. Launching an L2, deploying a set of smart contracts, launching a foundational DeFi module—all these can now be done in days, with engineering teams only needing a fraction of the resources they used to require. The impact on existing infrastructure is brutal: as supply approaches infinity, premiums evaporate. The more than 100 blockchains already live will be compressed to utility-level profit margins. Infrastructure that once commanded FDVs of $1 billion to $5 billion will be repriced based on what it actually earns. The venture capital playbook of “invest in infrastructure, capture narrative premiums, and exit to retail” is structurally broken.

But AI has also done something completely different, which is where crypto becomes essential.

When AI agents can generate infinite transactions, content, identities, and interactions, the cost of forging anything approaches zero. It becomes impossible to distinguish between spam and valid signals, between bot activity and human activity. In a world filled with limitless machine-generated noise, the only way to establish trust is through cryptographic proof.

Zero-knowledge technology has transformed from a niche scalability solution into essential infrastructure. Privacy-preserving credentials have shifted from academic research to the authentication layer of every AI-involved system. If you cannot prove who you are, that your transactions are authorized, and that your agent is qualified without exposing underlying data, you cannot participate.

This is the most underestimated proposition in the current crypto space: ZK and privacy technologies are not just privacy concepts; they are the trust layer of the AI economy.

Convergence of Two Forces: Tokens as the Operating Layer for Machines

Where institutions and AI converge, I believe the deepest opportunities lie.

AI agents have started trading autonomously. Coinbase just launched a wallet designed specifically for AI agents. The x402 protocol enables machine-to-machine payments. Autonomous systems are beginning to hold assets, execute trades, pay for compute, and interact with financial services without human intervention.

These agents need three things to operate.

Identity. Not a username—but a cryptographic credential that counterparties can verify in milliseconds without seeing the underlying data. Who does this agent represent? In which jurisdiction does it operate? What is it authorized to do? If you cannot answer these questions with programmable proofs, you will have to revert to centralized databases and manual reviews. This method cannot scale when millions of agents are trading simultaneously.

Programmable Assets. Stablecoins have proven that currency can be programmable and instantaneously settled. The same logic extends to government bonds, stocks, credit, and structured products. Agents do not care whether they hold USDC or tokenized government bonds. They care that the terms are readable, the rules are programmable, and the settlements are deterministic.

Proofs. Today’s compliance exists in human judgment and legal documents. It needs to encode regulation into machine-verifiable proofs—KYC status, licenses, jurisdictional authority, risk limits. The only way to deliver these at scale is through cryptographic proof, not centralized APIs.

This is where I have described the true collision of the two transformations. Institutions are pushing financial assets onto the chain because programmable infrastructure reduces settlement risks and operational friction. AI is pushing economic activities toward autonomous execution. When these two forces meet, the financial objects themselves must transform into software.

Machines will not buy tokens to speculate. They consume tokens to operate. This creates a demand curve that is entirely different from anything in crypto history. Retail speculation is cyclical, with narrative-driven capital rotating. Machine consumption is tied to the scale of autonomous economic activities. As AI systems automate more decision-making, trading, procurement, and coordination, the demand for machine-readable financial objects will expand in sync.

What This Means for Our Investments

I am writing this not to perform abstract analysis. This is the direction my new fund is actively investing in.

The old crypto game was: find the narrative, get ahead with tokens, and exit before unlocking. The new game is: identify the licensed infrastructure layer that captures stablecoin flows, build machine-readable primitives necessary for agent trading, and invest in jurisdictions where regulatory frameworks are taking shape first.

The companies that will win in the next era look and define entirely differently from the projects that defined the last era. They hold licenses, not just liquidity. They have revenue, not just TVL. They have regulatory moats, not just network effects. By the standards of those who speculate on memecoins, they might seem boring. By the standards of those allocating institutional capital, they present opportunities for a generation.

The era of token speculation gave crypto its start. Institutions and AI will give it its future. The transition between the two eras is happening right now and faster than most participants realize.

The speed of this transformation is a story that has not yet been fully told. Consider this a first draft.