Cathie Wood Says AI Agent Spending Could Reveal Where Demand Is Heading
Cathie Wood suggests investors monitor spending by autonomous artificial intelligence agents as a new indicator of technology demand and market traction, potentially reaching trillions by 2030.
Cathie Wood believes investors need to begin watching a new source of demand indicators soon: spending by artificial intelligence agents.
During her presentation at Robinhood’s Summit in Houston, the ARK Invest CEO suggested that market participants should focus on expenditures made by autonomous agents. Wood has updated her traditional maxim of “follow the developers”—which focused on spotting value by observing who builds useful technology—proposing instead that investors can increasingly “follow the agents” to determine which technologies are gaining real traction.
Related: XRP Scarcity Index Turns Negative: What Does It Mean for XRP Price?
Background: What Cathie Wood Means by “Follow the Agents”
The underlying concept of “follow the agents” is straightforward.
At present, investors seeking to spot emerging technology adoption typically analyze user metrics.
A growing user base, active developers, and rising download volumes or revenues all serve as indicators of mainstream acceptance.
Autonomous agents could soon produce a comparable signal.
An agent might need to spend funds to execute its programmed instructions, whether that involves paying for an API call, purchasing a good or service, or even compensating another software agent for completing a task.
If millions of agents start utilizing the same third-party API to access a specific dataset, that action indicates high data utility.
This is why AI agent spending holds potential value for investors: it elevates software adoption to an entirely new tier.
It translates adoption directly into financial transactions.
Why AI Agent Spending Could Become a Demand Signal
Among all economic metrics, cash remains one of the hardest to fabricate.
If autonomous agents consistently spend real money on a service, that expenditure carries clear value for the provider.
Consider a scenario where a research agent requires market data to conduct its analysis.
It faces a choice among three providers, including one it has utilized previously.
When the agent renews its API key, it makes a definitive statement regarding value: identifying the most dependable data source relative to its price.
Transactions of this nature can prove valuable to investors because they highlight which companies successfully capture demand from autonomous systems.
While these patterns may prove more difficult to forecast, they could offer deeper insights into long-term valuation trends.
That, according to Wood, is why investors should pay attention to “following the agents.”
Autonomous Agents Are Already Being Used To Make Purchases
Although the concept of automated payments may seem distant, enterprises are actively developing methods enabling autonomous agents to execute live transactions.
In July, Visa announced that its European agents were already testing autonomous transactions within controlled, real-world environments using live payment cards and merchants.
Amazon has also pursued this concept, announcing in August 2026 that Amazon Bedrock AgentCore Payments had achieved general availability.
This utility allows agents to discover and pay for APIs, Model Context Protocol services, and digital content independently of human intervention.
Related: Trump Could Put AI Under a 10-Member Watchdog as US Renames It “Super Intelligence”
Once this capability gains widespread adoption, AI agent spending will emerge as a significant economic driver.
How Big Could This Get?
ARK estimates indicate that agents could consume over $8 trillion in online goods and services by 2030.
Furthermore, ARK projects that their share of total digital spending could climb from approximately 2% in 2025 to 25% by 2030, arguing that agents can radically streamline the purchasing cycle and accelerate the journey from product discovery to final payment.
Rather than requiring a human buyer to execute each phase of a purchase, an agent could manage the entire workflow—ranging from product searches and price comparisons to checkout completion.
Naturally, none of these outcomes are guaranteed.
Even so, if even 1% of ARK’s projected value materializes, it would present an extraordinary opportunity for investors.
That explains why market participants should monitor AI agent spending closely.
What Will Autonomous Agents Spend Their Money On?
Software represents another category that agents will consume with much greater frequency.
A sophisticated agent may need to rent or buy access to multiple AI models or invoke an external API to execute tasks outside its core programming.
The same agent might likewise purchase datasets, rent compute power, or compensate another agent to carry out specific analyses.
For instance, an autonomous financial research agent might need to:
- Subscribe to a market data API to evaluate prices
- Pay another firm to inspect blockchain wallet activity
- Purchase access to a specialized model
- Scan a news repository for relevant stories
- Compensate a payment processor to settle each transfer
This dynamic introduces clear implications for traditional software sales models.
If agents can pay per query, per token, or per completed job, those pricing mechanisms could capture a significantly larger market share.
Conversely, software relying on seat-based or annual licensing models will hold far less appeal for autonomous agents.
Why Crypto Is a Good Fit for Agent Payments
Autonomous agents function best with payment networks capable of operating programmatically without demanding human authorization for every individual transfer.
This requirement makes stablecoins and alternative blockchain-native financial rails uniquely attractive for agent transactions.
Coinbase’s x402 protocol, for instance, revives the HTTP “402 Payment Required” error code by allowing computers to request immediate payment when encountering monetized APIs.
Amazon Bedrock AgentCore Payments already incorporates support for x402 alongside the Machine Payments Protocol (MPP). Concurrently, Coinbase’s x402 iteration enables agents to settle charges using USDC across blockchain networks like Base and Solana.
Should the population of autonomous agents skyrocket, machine-to-machine transactions will scale accordingly, with a substantial portion settling via stablecoins.
Of course, that shift is not guaranteed to occur.
It remains entirely possible that legacy card networks will maintain their dominance within the payments sector, even regarding autonomous agent activity.
Both Visa and Stripe are actively building proprietary tools designed to let agents authorize transactions using tokens familiar to human consumers.
Additionally, both payment processors have forged partnerships with OpenAI to facilitate payments utilizing models such as GPT.
Stripe has already rolled out several utilities permitting agents to disburse funds on behalf of human users.
These developments indicate that enterprises are preparing for the arrival of agentic commerce, even as crypto startups race to capture the lion’s share of the space.
Ultimately, the question is not a simple contest between card networks and stablecoins; rather, it centers on which payment infrastructure successfully attracts the largest volume of autonomous agents.
What Metrics Should Crypto Investors Watch?
While transaction volume will matter, it is unlikely to tell the entire story.
Investors will likely find behavioral patterns emerging among autonomous agents far more revealing.
For cryptocurrency markets, relevant metrics could encompass several dimensions:
- The volume of stablecoins expended by agents, the frequency of x402 payments processed, and the count of APIs opened to machine access
- Average transaction sizes and whether agents repeatedly pay for identical goods or services
- The specific blockchains utilized for agent transactions alongside associated fee structures
- The number of agent wallets generated and how spending is distributed across them
The utility of AI agent spending as an economic gauge will ultimately rely on how closely it mirrors true value generation.
An agent executing one million free API calls provides an intriguing data point, but it carries far less significance than a single agent making recurring payments for a specific service.
Therefore, while general agent activity is worth monitoring, AI agent spending holds far greater potential value for investors.
Related: XRP 2016 Pattern Returns: EGRAG Crypto Says $50 Could Be in Play
Could AI Agents Influence the Winning Blockchains?
If agents direct their time and capital autonomously, the outcome could reshape the broader technology sector.
Human consumers frequently select products and services influenced by brand reputation, marketing, or general availability.
Conversely, an autonomous agent can optimize its purchasing decisions based strictly on raw value.
It can favor one cloud computing infrastructure over another due to pricing or reliability, or select a database according to superior functional capabilities.
This autonomy allows agents to be much more selective about their chosen services, potentially shifting the balance of power among competing providers.
A blockchain network, for example, must deliver superior value to an agent compared to rival chains, or the software will simply migrate elsewhere.
This dynamic illustrates why AI agent spending could transform the competitive landscape of any industry capable of attracting autonomous systems.
Is AI Agent Spending a Useful Signal For Now?
Not yet.
Although clear long-term potential exists, AI agent spending remains in its nascent, fragmented stages—even if it is no longer merely hypothetical.
Many of the mechanisms ARK envisions that will drive massive surges in AI agent spending require further technical maturation.
Furthermore, much of the spending occurring today remains controlled by human developers rather than autonomous agents.
If an agent makes a purchase because a developer hard-coded that choice into the software, the transaction may not reflect genuine, independent value as clearly as investors might hope.
Why Does Cathie Wood’s “Follow the Agents” Matter To Crypto Investors?
The true appeal lies within a third category: potential value discovery.
If software agents begin adopting various technologies independently, that increased adoption could represent a significant source of value for the underlying technology providers.
This principle could apply across virtually any business technology category, ranging from databases and software applications to AI models and blockchain networks.
For the cryptocurrency sector, this opportunity is especially compelling.
Stablecoins and blockchain networks currently cater primarily to human-driven activity, but autonomous agents are already beginning to leverage them for machine-to-machine settlements. A sharp rise in agent payment volume could dramatically elevate the fundamental utility and value of blockchains.
Related: BitGo CEO Mike Belshe Warns US Crypto Is at Risk After CLARITY Act Failure
Whether that scenario materializes remains impossible to predict with certainty, but if it does occur, it will be driven by tangible utility rather than marketing hype or market narratives.
If Cathie Wood’s thesis holds true, investors can follow the money to locate those opportunities.
FAQ
01What Is AI Agent Spending?
AI agent spending encompasses financial transactions initiated or executed by autonomous AI systems while fulfilling tasks, including payments for consumer products, APIs, data sets, compute power, and additional digital services.
02Why Does Cathie Wood Think AI Agent Spending Matters?
Wood believes investors can increasingly “follow the agents” to identify which technologies autonomous software systems actively select and fund.
03How Could AI Agent Spending Benefit Crypto?
Stablecoins and blockchain platforms can facilitate automated, programmable machine payments without demanding human sign-off for every single transaction.
04Will AI Agents Use Stablecoins Instead of Cards?
Not necessarily. Legacy payment giants like Visa and Stripe are likewise designing payment infrastructure specifically tailored for autonomous agents.
05Is AI Agent Spending Already a Useful Investment Signal?
Not at this stage. Much of today’s agent-related expenditure remains human-directed by developers, meaning it does not always represent an autonomous choice among competing products or services.
