
Pantera Partner: AI Agents Are Becoming Customers—Who Will Secure First-Mover Advantage in Wallets, Compute, and Identity?
Author: Paul Veradittakit
Compiled by: TechFlow
TechFlow Summary: As AI agents begin to possess identity, budgets, and settlement capabilities, they are no longer just tools—they are becoming new customers. Paul Veradittakit maps out the real-world intersection of AI and blockchain today across four layers: currency settlement, credit capital, identity control, and compute sovereignty, highlighting projects already positioned at these intersections. For investors and practitioners focused on the agent economy and crypto infrastructure, this article serves as a practical reference for assessing the competitive landscape.

Summary:
- Last month I wrote that when prices stop rising, founder-market fit continues to compound. This month, Franklin Bi’s "The Battle for 8 Billion Customers" charts the target market we need to build. It encompasses 8 billion people, several agents per person, and enterprise agent fleets handling programming, finance, procurement, and sales.
- The product thesis is straightforward: wallets with limits, stablecoin settlements machines can execute, verifiable credentials agents can present, compute you actually own, and zero-knowledge proofs without public ledgers.
- Pantera already has a portfolio of companies building across each layer of this intersection, including: currency and machine settlement (Circle, Coinflow, OpenFX); credit, capital, and trading (Morpho, Ondo); identity, credentials, and control (World, TransCrypts, Alchemy); and compute sovereignty and real-time attestations (B3IQ, Orthogonal, Accountable).
- Founders and investors frequently ask me what is actually happening at the AI x Blockchain intersection. I plan to start writing about it here. If you are building at any of the layers mentioned in this issue, reach out to me or the Pantera team.
In "Founder-Market Fit," I argued that the pairing of a specific founder with a specific market is the only thing that can sustain compounding returns through a winter. Simply put, it’s depth, initiative, network effects, and obsession. Examples I’ve cited include Offchain Labs, Ondo, Morpho, Circle, and Alchemy.
Last week, my colleague Franklin Bi wrote an important piece, "The Battle for 8 Billion Customers," outlining the market founders are entering. What we are now facing and building for is 8 billion people, several agents per person, and enterprise agent fleets managing programming, finance, procurement, and sales. This translates to tens of billions of new decision and transaction endpoints. When agents possess identity and memory, budget permissions, selection and settlement capabilities, and accountability to their owners, they become customers. Gartner predicts agent influence on procurement will reach $30 trillion by 2030. Visa, Mastercard, and Coinbase’s x402 are already issuing credentials, enabling sub-cent payments. Cloudflare has also noted that bots now account for over half of HTTP requests.
The 8 Billion Customers
Eight billion people, each running a suite of personal and enterprise AI agents for programming, finance, procurement, sales, and logistics. This represents tens of billions of new independent decision and transaction endpoints.

AI models become economic entities the moment they possess three things:
Identity and Memory: Cryptographic credentials anchored to the entity.
Budget Permissions: Programmable limits, rate limits, and session keys.
Autonomous Settlement: The ability to discover, evaluate, and pay for services on-chain.
Franklin posed a critical question to builders: Who owns these agents? Is it the humans and enterprises they represent, or the platforms they run on?
If centralized cloud vendors own your agent’s identity, memory, and learning loops, switching providers means firing your digital workforce and starting over with a group of amnesiacs. Blockchain provides underlying property rights: portable identity, bounded delegation, and a settlement mechanism no model vendor can rewrite.
What AI x Blockchain Looks Like in Practice
1. Currency and Machine Settlement
AI agents won’t fill out KYC forms, wait three days for ACH transfers, or manage monthly credit card subscriptions. They need frictionless, sub-cent, 24/7 payment rails.
Coinflow: Seamlessly settles card and bank payments into stablecoins across 170+ countries, without exposing the underlying chain to users.
OpenFX: Processes hundreds of billions in annualized stablecoin settlements, explicitly built for software clients rather than human end-users.
Circle (USDC): Launched during the last bear market, it has now become the default unit of account for machine-to-machine microtransactions and enterprise agent settlement.
2. Credit, Capital, and Trading
When agents need to borrow or deploy capital based on configuration authorization, they require a programmable liquidity layer.
Morpho: Operates as a credit backend embedded by Coinbase, Robinhood, Societe Generale, and Apollo. It provides the default lending infrastructure for agents to programmatically query and borrow.
Ondo Finance: Converts tokenized US Treasuries and equities into productive, yield-bearing collateral. Nathan Allman left Goldman Sachs to tackle institutional asset tokenization, a problem that now directly intersects with agent budgeting.
FalconX (via acquisition of bloXroute): Assembles high-speed prime brokerage and execution stacks to serve always-on markets—the only operational timeline AI agents recognize.
3. Identity, Credentials, and Control
In a digital environment saturated with synthetic content, proving human intent and agent authorization is paramount.
World: Establishes foundational proof-of-personhood primitives; uses cryptography to prove the existence of a unique human actor in production environments, countering bot networks and sybil attacks.
TransCrypts: Puts employment, education, and legal credentials on user-controlled rails, allowing agents to verify permission claims without handing over sensitive underlying data.
Alchemy: Provides core developer platforms supporting agent-specific wallets and session key patterns. Developers no longer hand over master private keys to agents; instead, they issue fine-grained limits, including counterparty restrictions, expiration timestamps, and instant revocation capabilities.
4. Compute Sovereignty and Real-Time Attestations
B3IQ: Offers sovereignty-as-a-service through a buy-back compute network. To maintain genuine agent capability, institutions must ensure model weights and execution traces remain outside locked-in vendor environments.
Orthogonal: Functions as an orchestration and discovery layer for agent services. It is a leading platform offering metered access and native billing within decentralized networks.
Accountable: Enables financial institutions and autonomous funds to cryptographically prove real-time solvency without disclosing private balance sheets.
What We Look for in Category-Defining Founders
Sovereignty alone isn't enough to constitute a value proposition. Winning products leverage decentralized infrastructure to deliver lower transaction costs, stricter privacy guarantees, faster customization, or execution reliability that closed platforms cannot match.
When evaluating teams building at this intersection, we look for four traits:
Deep Domain Expertise: You’ve been inside the problem, not just read about it. (e.g., Ed Felten leaving Princeton and the White House to build Offchain Labs / Arbitrum).
Strong Initiative: You see market structures clearly enough that major institutions build around your tracks. (e.g., Paul Frambot founding Morpho at twenty to build the default credit engine for DeFi).
Unfair Network Effects: Distribution partnerships give you room to keep shipping during market downturns. (e.g., Jeremy Allaire partnering Circle with Coinbase to turn USDC into the global settlement standard).
Obsession: Belief in cross-cycle building even when attention shifts elsewhere. (e.g., Nikil Viswanathan and Joe Lau turning Alchemy into the default developer platform for Web3).
If you currently work at Goldman Sachs, Citadel, Stripe, Block, or a frontier AI lab: Technical skills that previously seemed adjacent to digital assets are now precisely the job descriptions needed to build an 800-billion-customer economy.
If you are building the core layer for the agent economy, reach out to me or the Pantera Capital team. We look forward to building with you.
Business
Bitcoin Posts Best August Since 2017
Bitcoin traded near $78,000, gaining approximately 24% for the month, marking its strongest August in nine years. Near the weekend, it briefly broke above $81,000. US spot Bitcoin ETFs absorbed roughly $3.3 billion in August. This marks the best month since October 2025 and temporarily pushed the asset class back above the $100 billion mark. This rally is being led by regulated US demand flowing back through the front door.
Tokenized Spot Equities Surpass $2.5 Billion
Tokenized spot equities crossed the $2.5 billion mark in August, up ~8% month-over-month and over 260% year-to-date. BNB Chain, Ethereum, and Solana split the market. DTCC is also conducting live simulations with ~40 firms. Participants include JPMorgan, Goldman Sachs, Invesco, and Citadel, utilizing tokenized equities and Treasuries as collateral. On-chain equities are beginning to clear like collateral, moving beyond mere wrapping.
USDC Supply Jumps, Circle Reclaims Transaction Share
USDC supply increased by roughly $2 billion over seven days after stagnating for six months. On-chain trackers also showed billions in minting activity that week, pushing circulating supply near $74 billion. Bernstein maintained an "outperform" rating on Circle with a $140 price target. The firm noted that USDC’s share of adjusted transaction volume has risen above 60% in 2026. Dollar tokens operating within the boundaries of US banking are exactly what institutions are expanding into.
Regulation
Senate Set to Vote on CLARITY Act September 15
Majority Leader Thune filed a motion for cloture to advance the bill. Following August’s recess, procedural voting is scheduled for September 15. The House has already passed the bill; the Senate’s window before the next recess is narrow, with ethics, DeFi, and Democratic support remaining key variables. Market structure will either receive codified law this fall or default to agency rulemaking for the remainder of this cycle.
SEC Proposes Crypto Asset Regulatory Framework Amid Legislative Standstill
On August 18, the SEC proposed a customized issuance regime: a $5 million startup exemption, financing tiers up to $75 million, conditional Howey safe harbors, and state registration primacy. Chairman Atkins explicitly stated that the proposal cannot replace legislation. When Congress pauses, agencies continue drafting market rules.
Treasury Proposes Rules Defining Stablecoin Issuance and Sale
The US Treasury released Rule III regarding the issuance of payment stablecoins and US offers, establishing a 60-day comment period. The statutory deadline for the final rule previously set for July 18 has been missed. The OCC now targets completion before November; the fallback effective date under the law is January 18, 2027. The framework is no longer a debate. It is a compressed implementation clock.
New Products and Trades
H1 2026 Crypto M&A Hits Record High
Mastercard completed its acquisition of BVNK on August 3, integrating the stablecoin payment stack into its global card network. Crypto M&A total reached a record $9.7 billion in H1 despite a decline in deal counts. Massive capital is flowing toward licensed payments rather than permissionless experiments. Buyers in the digital asset space are now established payment incumbents.
Coinbase Establishes Tokenization Hub in Abu Dhabi
ADGM granted Coinbase a license to arrange investments and custody tokenized securities backed by underlying equities. This authorization expands Project Diamond into a regional issuance and custody hub. Transfers remain within a framework featuring sanctions screening and wallet-level controls. Tokenized equities are moving past demo phases and securing regulated footholds outside the US listing queue.
BlackRock Adds Tokenized Money Market Products Built for GENIUS Reserves
BlackRock launched an on-chain share class for a Treasury liquidity fund, along with a multi-chain daily reinvestment reserve tool. Both aim to meet qualified reserve requirements for permitted US payment stablecoin issuers. The firm already manages roughly $60 billion in Circle reserves and aims to become the default reserve manager for this category. Stablecoin growth is increasingly becoming an asset management exercise.
Pantera News
Positions Shift from Shorts to Longs After Bitcoin Recovers 200-Day Moving Average
Pantera General Partner Cosmo Jiang noted that since the October 2025 peak near $126,000, Bitcoin has endured a roughly ten-month, ~50% drawdown. With Bitcoin holding above the ~$69,000 200-day moving average, traders are rotating from net shorts or cash into long exposure. He pointed out that more favorable US policy and larger-scale Treasury buybacks were direct catalysts, with $80,000 as the next resistance level.
Fortune Reports on B3IQ's Buyback Compute Model
Pantera portfolio company B3 Labs launched B3IQ. Researchers and enterprises can purchase US-hosted NVIDIA machines with a 30% down payment instead of leasing hyperscale compute locked until 2030. Early adoption has moved quickly; the team reported eight-figure GPU sales within six days and is already scaling capacity for startups, labs, and data-sensitive industries. Crypto-native operators are becoming the actual suppliers of scarce compute.
On Stateful: Why AI Builders Need to Own Their Own Compute
Watch Show · Listen on Stateful
Franklin Bi sat down with B3 Co-founder Daryl Xu, Viktoriya Hying, and NYU’s Yorke Rhodes to discuss the launch. The cost of leasing an H200 node for two years roughly equals outright purchasing. Idle periods are matched with buyers, allowing the hardware to pay for itself. Closed models have already barred research teams from working on human trafficking and conflict-zone evacuation tasks. Open-source models are catching up to the frontier. Owning the hardware is the unlock that makes them truly accessible.














