Cathie Wood Says Investors May Need to ‘Follow the Agents’ in AI Era
Cathie Wood Says ‘Follow the Agents’ as AI Payments Become Crypto’s Next Adoption Test
Cathie Wood is adding a new indicator to the technology investor’s playbook: watch where artificial intelligence agents spend money.
“We’re probably going to be talking more and more about ‘follow the agents,’” the ARK Invest CEO said during a panel at Robinhood’s Summit in Houston, adapting her longstanding advice that investors should “follow the developers” to identify technologies gaining adoption.
The distinction matters as AI moves from generating information to acting on users’ behalf. Autonomous agents can increasingly select software, purchase data, access computing resources and execute financial transactions, potentially turning their choices into a measurable source of demand for technology platforms and payment networks.
For crypto investors, the question is whether stablecoins and blockchains become the financial infrastructure those agents prefer, or whether most machine-driven commerce stays on card networks and other conventional payment systems.
What Does It Mean to ‘Follow the Agents’?
Developers have historically provided an early indicator of technology adoption because their decisions reveal which programming languages, cloud platforms and networks are attracting builders before mass-market demand becomes obvious.
AI agents could create a different signal. An autonomous system searching for market data, buying computing capacity or booking travel may evaluate competing services repeatedly and at machine speed. If enough agents begin making those decisions, their transaction patterns could reveal which digital infrastructure is winning real usage.
That activity has already started appearing on blockchains. FinanceFeeds reported in August that x402 recorded 8.7 million machine-to-machine transfers in a single week. Yet those payments carried only about $368,000 in value, or roughly four cents per transaction on average.
The figures illustrate both sides of the investment thesis. Agents can generate enormous transaction counts, but current activity is still dominated by low-value API queries, data requests and other micropayments rather than meaningful consumer commerce.
Investor Takeaway
Agent transaction counts may become a useful adoption metric, but investors will need to distinguish activity from economic value. Rising settlement volume and revenue generated from agent transactions would provide stronger evidence than transaction numbers alone.
Why Could Stablecoins Fit Machine-to-Machine Payments?
AI agents create payment requirements that differ from ordinary consumer spending. Software may need to purchase one database query, an API call or a few seconds of computing capacity without creating an account, entering card details or waiting for a person to authorize every transaction.
BlackRock has identified that overlap between AI and digital assets as a potential adoption driver. The asset manager’s 2026 thematic work describes AI as machine-native intelligence, crypto as potential machine-native money and blockchains as programmable infrastructure connecting the two. FinanceFeeds previously examined BlackRock’s case for blockchains as settlement infrastructure for autonomous agents.
Coinbase’s x402 protocol is one attempt to build those rails. It allows software to encounter a payment request, settle it programmatically and continue its task without a conventional checkout process. Coinbase has expanded the same infrastructure into autonomous trading, while CEO Brian Armstrong has said Grok is currently the leading client for agentic traders on Coinbase, without providing activity figures.
The infrastructure is also spreading beyond crypto-native companies. FinanceFeeds reported that Visa, Mastercard, Ripple and other major payments and technology companies backed the x402 standard, increasing the possibility that agentic payments develop around interoperable infrastructure rather than one company’s closed network.
Investor Takeaway
The investment opportunity may sit less in AI agents holding speculative crypto assets and more in stablecoins, settlement networks and payment protocols that agents repeatedly use to buy digital services.
Will Crypto Actually Win the Agentic Payments Market?
That outcome is far from settled. Joseph Chalom, co-CEO of SharpLink and BlackRock’s former head of digital assets, has argued that agentic finance should operate on open infrastructure rather than become controlled by a small group of banks or technology companies.
His framework emphasizes limited spending authority, revocable permissions, transparent transaction records and portability between financial providers. A user might allow an agent to spend $500 booking a hotel, for example, without giving it unrestricted access to an entire account.
Open blockchains could provide common infrastructure for that model, but crypto faces strong competition. Visa, Stripe, Google and other payments and technology companies are developing their own agentic-commerce systems. FinanceFeeds has also reported that Visa expects cards and stablecoins to coexist in AI commerce, with conventional rails potentially retaining consumer purchases while stablecoins handle high-frequency machine-to-machine transfers.
That makes Wood’s “follow the agents” concept particularly useful for investors. Rather than assuming blockchain will become the default payment layer for autonomous software, markets may eventually be able to observe where agents actually transact, which assets they use for settlement and which networks capture the resulting fees.
Investor Takeaway
The important metric will be where agent spending ultimately settles. If stablecoins begin capturing commercially meaningful volumes rather than primarily micropayments, AI could become a new source of structural blockchain demand. If agents remain concentrated on cards and closed payment platforms, the crypto opportunity will be narrower.
