TechFlow Logo
Login/ Sign up
ETH Gas
Gwei
Fear
gas
BlackRock Proposes "Machine-Native Economy": Which Chains and Tokens Can Generate Revenue from Agent Payments?

BlackRock Proposes "Machine-Native Economy": Which Chains and Tokens Can Generate Revenue from Agent Payments?

2026.09.24
Share
For chains and tokens, competitors are not just other chains, but also the agent-driven upgrade of the entire traditional payment system.

By Xiaobing

Global asset manager BlackRock released a research report this week titled "The Machine-Native Economy: How Digital Assets Connect Intelligence, Commerce, and Compute," proposing a concise and bold equation: AI is "machine-native intelligence," digital assets are "machine-native money," and blockchain is the programmable infrastructure connecting the two.

The report was primarily written by BlackRock's Digital Asset Research team, led by Will Su and Robert Mitchnick, who also heads BlackRock's iShares product line.

The core thesis of the report is: as more software agents (AI Agents) begin autonomously purchasing data, calling APIs, and renting compute, they require a payment rail that does not rely on humans clicking "confirm." Traditional bank accounts and credit card systems cannot serve software without IDs, while stablecoins and blockchain perfectly fill this gap.

In plain terms: robots need to spend money, but banks won't open accounts for them, so robots have to use cryptocurrency.

This narrative isn't new in the crypto industry, but having an institution of BlackRock's caliber back it carries different weight.

The key question is: how far is the path from narrative to revenue? Which chains and tokens truly stand on this path, and how many are merely bystanders along the way?

The Report Paints Three Big Pictures

BlackRock's framework consists of three layers: stablecoins as "tools for spending," Bitcoin as "tools for saving," and tokenized compute as a "new asset class."

Stablecoins: The only layer with genuine traffic. As of September 2026, the total market cap of stablecoins exceeded $300 billion, with adjusted trading volume surpassing $11 trillion for the full year of 2025, roughly matching Visa's processing volume during the same period. USDT leads with approximately $189.5 billion, followed by USDC at around $77.3 billion. This is already operational infrastructure, not a dotted line on a roadmap.

Visa's stablecoin settlement project has processed $7 billion annually across nine chains. Mastercard has also integrated eight chains, including Solana, Arbitrum, Base, Ethereum, and Polygon, into its stablecoin settlement network. SoFi has just become the first U.S. national bank to launch on Mastercard's stablecoin rails, deploying 70% of its SoFiUSD supply on Solana.

Bitcoin: Beautiful narrative, but the evidence is shifting. BlackRock cited a study from earlier this year by the Bitcoin Policy Institute, which tested 36 AI models in simulated scenarios and found they chose Bitcoin as a store of value in 79.1% of cases, and stablecoins for payment in 53.2%. While this data appears highly favorable to Bitcoin, the latest round of tests (early September, using 9 updated models) showed a clear shift: the proportion choosing stablecoins for payment rose to 77%, while the proportion choosing Bitcoin for payment dropped from 36% to 20%. Only Grok 4.6 lists Bitcoin as its preferred payment tool. BlackRock's report doesn't even mention Bitcoin's Lightning Network or the L402 protocol; instead, it references Coinbase-led x402, Stripe's MPP, Google's AP2, and Visa's TAP.

From another angle: BlackRock assigns Bitcoin the role of a "savings account," but in the actual payment processes of the Agent economy, Bitcoin has already been marginalized.

Tokenized Compute: Highest imagination, furthest from reality. The most striking forward-looking assessment in the report is: "As Agents become more persistent and powerful, standardized claims on compute capacity could become a significant use case for digital assets in financing and programmable settlements." BlackRock estimates that the combined revenue of AWS, Microsoft Intelligent Cloud, and Google Cloud will reach approximately $1.1 trillion by 2030. If compute becomes a tradable standardized commodity like oil or wheat, the market space for its financial derivatives would clearly be much larger.

But this remains a theoretical possibility. No currently existing tokenized compute market possesses the level of standardization required for commodity futures.

$5,000: The Real Monthly Volume of Agent Payments

The report acknowledges that Agent payments are still early-stage, but just how early is this "early"? A study released on September 9 by blockchain intelligence firm TRM Labs provides an incredibly sobering answer.

Since its launch in May 2025, the Coinbase-led x402 protocol has accumulated over 205 million settlement transactions totaling approximately $52.7 million. On the surface, these numbers seem exciting. However, TRM Labs applied a three-layer filter: removing self-payments where the payer and seller addresses match, eliminating abnormal traffic concentrated among a few payers, and excluding sellers with fewer than 10 buyers. After filtering, roughly half of the transaction volume evaporated, leaving approximately $25.62 million that can be classified as legitimate buy-sell transactions.

Of that $25.62 million, only 0.6% to 7.5% of transactions appear to originate from actual AI Agents. Converted to absolute terms: between $154,000 and $1.92 million. At a monthly run rate for 2026, true Agent commercial payments amount to roughly $5,000 to $11,000 per month.

$5,000 per month. This is the actual Agent trading volume generated by a protocol promoted by Coinbase, running across three chains, and supported by 100,000 registered Agents.

The primary signal TRM Labs uses to identify AI activity is transaction amount: trades averaging under $1 are more likely to stem from micro-payments by Agents. But the researchers themselves note that "dedicated Agents" designed for single tasks may not fit this pattern, meaning actual Agent activity could be underestimated. Even so, a gap ranging from thousands to hundreds of thousands of dollars is wide enough to make one question whether the entire market even exists.

USDC accounts for 99.6% of the settlement value on x402. This validates BlackRock's assertion that "stablecoins are the preferred payment tool for Agents," but it also means x402 currently functions more as a USDC conduit than a diversified Agent economy.

Who Is Actually Making Money: The Revenue Reality Across Settlement, Compute, and Data Layers

BlackRock's report outlines a layered system: the base settlement networks (L1/L2) collect gas fees from each transaction; compute and data markets in the mid-layer earn commissions from supply-demand matching; and stablecoin issuers at the top profit from interest on reserve assets. This framework is correct, but the revenue scale across each layer varies enormously.

Settlement Layer: Base Captures the Largest Initial Dividend. 67% of x402's volume runs on Coinbase's L2 network Base, with the rest distributed across Solana and Polygon. However, the gas revenue generated from Agent payments is negligible for these chains. Assuming an average of $0.30 per transaction and gas fees of about $0.001, the total volume of $52.7 million generates only tens of thousands of dollars in on-chain fees. For L1/L2s, a truly meaningful signal will only appear when the scale reaches billions of transactions per month. Currently, these chains earn far more from conventional stablecoin transfers and DeFi activities than from Agent payments.

Solana is the most active chain in stablecoin settlement. In April 2026, it captured 32.6% of the weekly adjusted stablecoin transfer volume, surpassing Ethereum's 27.8%. In February alone, stablecoin trading volume on Solana reached $650 billion. But this volume stems primarily from DeFi and cross-border payments, not the Agent economy. Directly attributing Solana's stablecoin processing volume to Agent payment dividends is a logical leap.

Compute Layer: Revenue is growing, but attribution is blurry. Decentralized compute markets are the most direct mapping of BlackRock's "tokenized compute" narrative. The actual revenue data for top projects is as follows:

Aethir is currently the highest-revenue DePIN compute project, generating $127.8 million in 2025, with annualized recurring revenue reaching $166 million in Q3, spanning 94 countries and over 440K GPU containers. This revenue comes from AI training, inference, and cloud gaming demands from more than 150 enterprise clients.

Aakash's data tells a more complex story. Aakash officially reported that compute expenditures across all periods surpassed $5 million for the first time in Q1 2026; however, independent tracking by Messari shows that on-chain leasing revenue for the same quarter was only $253,000, down 45% quarter-over-quarter, annualizing to about $1 million. The difference between the two figures is roughly fivefold, likely stemming from differing accounting methods for AakashML inference business revenue. AakashML's inference throughput did indeed grow from approximately 5 billion tokens/day in May to over 10 billion tokens/day in July, but the AKT token (currently around $0.51, down ~94% from its 2021 peak) failed to capture these gains.

io.net reported approximately $20 million in annualized on-chain revenue, boasting over 10,000 active nodes and 56 enterprise clients, costing roughly 72% less than AWS.

Render Network has accumulated over 77.5 million rendered frames, 5,600 nodes, and a token burn volume up 279% year-over-year, but protocol-level revenue data has never been publicly disclosed.

Aggregated, the DePIN compute sector tracked by DeFiLlama had an annualized revenue of roughly $180 million to $220 million in Q1 2026, with Aethir taking the lion's share. Yet the entire DePIN ecosystem, with a circulating market cap of approximately $10 billion, only generated $72 million in on-chain revenue during fiscal 2025. This is not an undervalued track; it's a track with a massive chasm between revenue and valuation.

A more critical question is: the growth in compute market revenue is currently driven by the natural spillover of enterprise AI demand (cloud providers facing capacity constraints, leading clients to decentralized alternatives), which has almost nothing to do with Agent payments. The scenario described by BlackRock of "Agents autonomously purchasing compute and settling with stablecoins" has not yet scaled on these platforms.

Data Layer: The most barren link. BlackRock's report lists "data purchasing" as one of the core activities of the Agent economy, but in the realm of on-chain data markets, there are virtually no revenue cases worth discussing. CoinGecko's paid API is one of the early deployments of x402, but the volume of such services is too small to support a standalone asset class narrative.

Competitive Landscape of Agent Payment Protocols: Crypto Is Not the Sole Player

The five Agent payment protocols listed in BlackRock's report reflect a fact selectively ignored by the crypto industry: the protocol competition for Agent payments is far more complex than "on-chain vs. off-chain."

x402 is a stablecoin payment protocol built by Coinbase around the HTTP 402 status code. The Agent initiates a request, the server returns a price and address, and the Agent settles on-chain with USDC without requiring an account. This is the crypto-native path.

MPP (Machine Payments Protocol), jointly developed by Stripe and Tempo, launched in March 2026 and is characterized by supporting both stablecoin and fiat settlements. Anthropic, OpenAI, Shopify, Etsy, Visa, and Mastercard are all signatories. Stripe's strategy is clear: before x402 can seize the high ground of standards, leverage distribution advantages to lock in norms.

ACP (Agentic Commerce Protocol), co-developed by OpenAI and Stripe, standardizes the checkout flow within AI interfaces. AP2 is Google's authorization layer protocol, addressing the question of "who approved this transaction." TAP is Visa's Trusted Agent Protocol, solving the identity verification problem of "is this Agent a legitimate buyer."

These protocols are mutually composable, forming a layered stack: a single Agent purchase might simultaneously use AP2 for authorization, ACP for checkout, and x402 or MPP for settlement. Crypto payments (x402) occupy only a portion of the settlement layer in this stack. If Stripe's MPP allows Agents to complete identical micro-payments using fiat, then the exclusivity of on-chain settlement disappears.

For chains and tokens, competitors aren't just other chains, but the agent-ification upgrade of the entire traditional payment system.

Where Is the Real Alpha?

To return to the opening question: which chains and tokens will Agent payments generate revenue for?

If we define "capable of generating revenue" as "already generating verifiable revenue from Agent activities today," the sobering answer is: almost none. The real Agent payment volume of $5,000 to $11,000 per month, distributed across Base, Solana, and Polygon, has zero impact on the revenue reports of any single chain.

If we extend the timeframe to "likely to benefit in the next 3-5 years," the rough ranking of beneficiaries is as follows:

Stablecoin issuers (Circle, Tether) are the most certain winners. Regardless of which chain Agents trade on or which protocol they use, USDC and USDT remain the default options for payment mediums. However, USDC and USDT themselves are not investable tokens, and Circle's revenue comes from reserve interest, not on-chain fees.

Low-fee settlement chains (Solana, Base) hold a structural advantage in the competition. Typical Agent transaction amounts fall below $1, making a $0.001 gas fee a prerequisite. The Ethereum mainnet is naturally excluded from micro-payment scenarios. However, Agent payment revenue on Solana and Base is currently rounded to zero.

Decentralized compute networks (Aethir, Akash, Render, io.net) have real revenue, but this income stems from enterprise AI demand, not autonomous Agent procurement. The "compute tokenization" scenario described by BlackRock, where Agents buy and sell standardized compute like futures contracts, does not exist on any platform today. The AKT token price has fallen 94% from its peak, and the market has not granted Akash usage growth a valuation premium.

The value capture mechanisms for protocol-specific tokens are generally weak. Most DePIN project tokens lack direct revenue distribution rights, their token burn volumes are negligible relative to market cap, and token inflation far exceeds demand absorption. Render's token burn volume grew 279% year-over-year, but cumulative burn volume accounted for only about 0.2% of its market cap by the end of 2025.