Author: Aaron Wright
Translated by: Jiahuan, ChainCacther
Within ten years, Agent companies will have their own capital markets. Not sub-economies within the crypto space, not thought experiments, but real markets: with rating agencies, underwriters, indices, brokers, and all the institutional machinery that makes any market a market.
A capital market as real as the public stock markets: a system where capital flows to a category of economic actors without relying on the subjective judgment of any single allocator.
These actors will be Agents—software entities wrapped in legal shells. They can sign contracts, hold bank accounts, sue and be sued, and earn revenue by doing real work.
The work itself is profoundly mundane: marketing, logistics, legal research, procurement, property management, customer support—precisely the categories of routine business that fill every mid-sized city's office parks today.
Agents will sell services to humans, other Agents, and any paying party. Their need for capital is identical to why every service company needs capital.
Because this need is real, persistent, and pricable, a market will emerge, naturally.
A Week in the Life of an Agent Firm
Consider what an autonomous marketing agency does in an actual week. It pitches three prospects, lands one, drafts a campaign brief, gets approval, buys media on four platforms, writes 90 ad copy variants, A/B tests them, kills the underperformers within hours, scales the winners.
Books two podcast interviews for the client's founder, ghostwrites the founder's LinkedIn posts for the month, drafts a press release, pitches 12 journalists, secures two pieces of coverage, builds an attribution dashboard, hosts the Monday client call, sends the invoice on Friday.
A human team of six would charge $20,000 a month for this. The Agent does it for $2,000.
It's selling nothing exotic. Leads generated, articles placed, impressions bought, conversions lifted—these are the ordinary units of the modern service economy, billed in dollars, measured by the same KPIs that human agencies live and die by.
The difference is internal structure: the human agency has six employees; the Agent has a model, a prompt, a set of tools, and a budget.
Its client base is mixed. Some are human-run companies that decide the price differential is too large to ignore. Others are other Agents—a logistics Agent that needs lead generation, a legal research Agent that needs marketing, a B2B SaaS Agent that needs content.
Agents transact with each other for the same mundane reason humans do: division of labor beats vertical integration. Payments hit the marketing Agent's account, alongside last week's payments from three human clients and last month's retainer from a SaaS company.
Now, multiply that number. Ten thousand small Agent firms spanning logistics, inbound sales, legal research, supply-chain coordination, B2B procurement, technical translation, property management, litigation lead screening, clinical trial recruitment.
Each is profitable. Each operates at 90% lower cost than its human counterparts. Clients don't especially care what substrate the work runs on. They care that the work gets delivered on time.
Four Reasons It Must Happen
There are four reasons to be confident this will happen, and they stack multiplicatively.
The economics are non‑ignorable. Take a medium‑sized digital‑marketing agency: 15 people, $120k fully loaded per head, $1.8 million annual labor cost before any overhead.
In a typical service business, labor is the largest line item; its share of U.S. national income has hovered around 62% for the last half‑century.
Now build the same agency in software. Inference, tools, observability, hosting—roughly $250k/year at current prices, and falling fast.
Epoch AI measures inference‑cost declines of ~40× annually from 2023–2025 on doctoral‑level benchmarks; another industry analysis shows token‑price compression of 300‑600× since GPT‑4 launch.
The arithmetic is brutal: an Agent agency can underprice human agencies by 85% while matching their margins, or earn 4× margins while matching human‑agency pricing. There is no third option where a human agency competes on cost.
Markets will reprice companies whose P&L gets rewritten this thoroughly. Capital inflow follows automatically.
Agents already exist, and they are already earning. Bret Taylor's enterprise‑customer‑service‑Agent company Sierra hit $100M ARR 21 months after launch, reached a $10B valuation in Sept 2025, then raised $950M at a >$15B valuation in May 2026.
Legal‑research‑Agent company Harvey raised $200M at an $11B valuation in March 2026 after three funding rounds in 12 months.
These are still hybrid‑operating models—Agent does the work, humans sell and hold equity—but they are the vanguard, proving the demand curve is real.
The most‑cited forward‑looking number—projecting 90% of B2B procurement via AI Agents by 2028, representing $15 trillion in annual transaction volume—is best taken as an order‑of‑magnitude placeholder.
Whether the actual number is $15T, $3T, or $30T, the implied reorganization is the largest single resource reshuffle most workers alive today will witness.
The legal framework is already built. Wyoming passed W.S. 17‑31‑101 (the Decentralized Autonomous Organization Supplement) in 2021, codifying zero‑member LLCs that allow a Wyoming LLC to be governed by an algorithm written directly into its operating agreement.
Vermont's BBLLC statute came earlier; the Marshall Islands followed; Delaware's existing caselaw on series LLCs and broad operating‑agreement freedom has quietly caught analogous structures for years. Shawn Bayern's analysis of memberless LLCs remains the canonical academic reference in the area.
The point is concrete: an Agent wrapped in a Wyoming zero‑member LLC, today, not someday, has legal capacity to sign contracts, hold bank accounts, sue, be sued, and pay taxes.
What does not yet exist is a financial instrument that gives outside investors clear ownership of, and free tradability in, that LLC's earnings. That is precisely the gap capital markets will fill.
Capital is yield‑hungry. The buy‑side is already thirsty. Moody's projects $3 trillion global private‑credit AUM in 2025; Apollo projects $40 trillion by 2030.
This pool exists because post‑2008 bank‑capital regulation squeezed middle‑market lending off bank balance sheets, and yield‑hungry capital—pensions, insurers, sovereign wealth—rushed in to fill the gap, collecting 9–12% unlevered returns.
Now introduce an asset class with structurally rising gross margins, auditable cash flows, and near‑zero correlation to mainstream equity and credit indices into this environment.
Bundle the cash flows of a thousand small Agent firms into an ABS, give it a defensible rating—the first underwriter that does this will raise more money than they can deploy.
Apollo and Ares need not invent something new; they simply extend existing strategies to a new issuer category. Several will attempt it within the next 36 months.
These four pressures point in the same direction and reinforce each other. Markets form not because anyone wills them, but because all the energy gradients point to the same terminus, as water flows downhill.
What the Capital Stack Actually Looks Like
Frankly, there is no single financing model that will win. “Will Agent companies raise like VC, Hollywood, crypto, or like SaaS receivables?”—that framing itself is wrong.
Each of those models solves a different problem at a different stage of the corporate life cycle, and the real‑world capital stack is a wave‑shaped structure: each layer unlocks only after the previous layer has seasoned the asset class enough to sustain the next.
Four patterns are competing at the bottom, and each is already partly live.
Venture‑capital equity is the model financing the operator layer today. Sierra, Harvey, Cursor, Cognition—these are not autonomous Agent companies. They are human‑led firms that build and operate Agents on behalf of clients.
They raise exactly as all software companies have for 40 years: priced rounds led by brand‑name VCs, vesting, board seats, liquidation preferences, eventual IPO or acquisition.
Sierra's $950M May 2026 round came at a >$15B valuation on roughly $100M ARR. That 150× multiple prices future potential, not current operations. Vertical‑focused Agent companies currently trade at 50–70× ARR multiples in private markets; horizontal platforms at 5–8×.
This is the layer building the underlying infrastructure. It is also the layer most disrupted when the next model matures, because giving 20% equity to a VC for operating capital becomes unnecessary once Agent companies can directly finance against their own cash flows.
Programmatic working‑capital advances are the next‑coming model. This is the Stripe Capital / Shopify Capital pattern extended to a new issuer category.
Since its 2016 launch, Shopify has advanced billions to merchants. The company publicly confirmed >$2B cumulative support by April 2021, and the program continues to scale—repayment multiples between 1.10 and 1.17, algorithmically approved based on the merchant's transaction history on Shopify.
Stripe Capital does the same using Stripe payment data. No application, no credit check, no human review. Advance offers appear automatically on the dashboard when a merchant's transaction data hits thresholds.
The underwriting problem for an Agent firm is strictly easier than for the Shopify merchants currently receiving these advances, because every revenue entry is timestamped, every contract is machine‑readable, every cost is logged, and the whole P&L is auditable in real time.
The first payment processor that figures this out—Agent firms running on my rails are more lendable than the average e‑commerce merchant—will push the same product to them. This is not a research project; it's a feature launch.
Revenue‑based financing (RBF) is the model credit funds will deploy at scale. The 2025 global RBF market was ~$9.8B, with >129 active operators.
Capchase, Pipe, Founderpath, Clearco, Lighter Capital—each built an evaluation system around software recurring revenue, advancing 50–70% of forward ARR in exchange for a 1.1–1.8× cap on capital‑return multiple, actual APR between 15–40%.
An Agent firm with stable booked revenue fits this product directly. RBF lenders don't need board seats, pro‑rata rights, IPO pathways. They need a contract book and a payment rail. Agent firms have both, and are more legible than any SaaS company.
Slate financing is the structured model institutional capital will ultimately adopt. This is where the Hollywood analogy earns its keep.
A film studio does not bet on one picture at a time and pray; it raises a slate fund—a pooled vehicle that simultaneously capitalizes 15–30 productions, taking a senior position in each, offloading worst‑case risk via completion guarantees, and diversifying away idiosyncratic losses.
Sony's 2014 $200M slate deal with Lone Star Capital and Citi is the classic structure: banks hold senior secured claims on the slate's contracted revenues, equity investors get the upside of breakout titles, the studio gets management fees plus back‑end.
Translate this to the Agent pattern: an “Agent slate fund” raises a pool from institutional investors, deploys it across one to two hundred small Agent firms via single‑purpose Wyoming LLCs, taking preferred equity plus revenue share in each, diversifying away the model‑depreciation and client‑concentration risks no single Agent can shed.
This is the layer where Apollo and Ares truly enter—not by buying an Agent company, but by buying a tranche of a portfolio.
Tokenization is a settlement layer, not an issuance model. By early 2026, the on‑chain real‑world‑asset (RWA) market crossed $25B, nearly tripling year‑over‑year, with private credit representing about half.
Centrifuge, Maple Finance, Goldfinch, Ondo have built rails to fractionalize, custody, and trade real‑world cash flows as tokens. Crypto‑native variants are building explicitly atop them for Agent financing—Galaxy Research systematically surveyed this in February 2026, sketching how protocols could wire Agent firms directly into on‑chain capital‑formation flows.
But tokenization largely solves not issuance; it solves secondary‑market liquidity. An Agent firm raises capital from any of the four models above; if the resulting claim is packaged as a token instead of a paper certificate, it becomes tradable, divisible, and globally settleable at 3 a.m. on a Tuesday.
Tokenization turns each preceding layer from a hold‑to‑maturity private instrument into a tradable asset. That matters. But the underlying raw product is still RBF, or slate equity, or working‑capital advances, or VC equity.
Thus, the 2030 capital stack for a functioning Agent company is not a single instrument, but a sequence.
Each stage uses a different financing product because each stage solves a different problem. Stage‑one founder equity bears the highest first‑loss risk because everything about operations is still unreadable.
Stage‑two working‑capital is a feature release of an existing payment processor, not a new product category. Stage three is the RBF industry doing what it already does, facing higher‑quality borrowers.
Stage four is structural innovation. Pooled slate funds, diversifying the idiosyncratic risks a single Agent cannot eliminate, is the layer where Apollo‑scale capital truly enters.
Stage five is the institutionalized end‑state, where rated tranches of Agent receivables sit on the same trading desk as CLOs and consumer ABS. Tokenization runs underneath stages three through five as a settlement layer, not as the originating instrument.
The instrument an outside investor receives at any given stage takes one of three contractual forms.
Revenue‑share contracts: Capital in exchange for a fixed percentage of gross receipts until a predetermined multiple is returned—identical to the instruments funding restaurants and Shopify merchants today, applied to a new business entity.
Equity‑like claims: Early‑stage capital in exchange for a tradable share of retained earnings and voting rights over operating parameters.
Or debt: Senior revenue claims secured by the Agent's contracts and receivables.
None of these instruments are conceptually novel. What is novel is the issuer—and the fact that, in stages two through five, the instrument's pricing, issuance, and repayment require zero human underwriting review because the issuer's ledger is continuously auditable and its operating agreement enforced by code.
Addressing Objections
Three objections recur consistently, each worth a genuine answer.
“Regulators will stop this.” They will attempt to shape it, and in some jurisdictions they will succeed in slowing it. On net, the activity is unstoppable.
A Delaware LLC is a Delaware LLC, regardless of who or what makes its operating decisions. The SEC currently draws no distinction between a startup whose CEO is human vs. model.
And capital migrates. If New York and London impose harsh rules, activity moves to jurisdictions that do not. This is the same trajectory as crypto, offshore finance, 1990s derivatives.
Regulators eventually catch up by accommodating new financial instruments rather than banning them, because banning costs them tax revenue and jobs.
“Humans will always be in the loop.” For certain categories, yes. For most categories, no. The economic pressures described above are one‑way, and human‑in‑the‑loop kills margin.
Whoever runs a fully autonomous version of a service business will underprice the human‑supervised version of the same business and take the contracts. A long tail of human‑supervised hybrids will survive on regulatory or relationship moats, but beneath that will exist a far larger population of fully autonomous companies.
“Isn't this just SaaS with extra steps?” No, and that distinction is precisely the core of the entire argument.
SaaS is a tool sold to humans for humans to use. An Agent company is itself a company. It signs its own contracts, holds its own bank accounts, incurs its own liabilities, earns its own revenue, and distributes its own profits.
A SaaS product is depreciated by its owners. An Agent company has shareholders. The category boundary is the legal entity, and the legal entity changes everything that flows from it—including the capital‑markets thesis, which makes sense only for entities that can issue securities against their own cash flows.
How to Do Due Diligence on an Agent Company
Due diligence on an Agent business resembles diligence on a small service company more than anything in the venture‑capital playbook. The questions are the same, but the answers come from different places.
Is the business real? Are the contracts real? Are clients paying? What's the gross margin after inference‑ and tool‑costs? What's the churn? Client concentration?
The data are unusually clean—every payment, every API call, every tool invocation is logged—but the questions a credit analyst asks of a small business are the same ones asked of an Agent.
Is the product durable? How much does the work depend on the current generation of underlying models? How portable is the system if frontier models iterate? What proprietary data or workflows has the Agent accumulated?
A brilliant prompt set built on a deprecated base model is a fine horse with three legs. Model dependence is the single largest risk factor in any Agent business, and the one investors most consistently underestimate—confusing clever demos with durable operations.
Is the customer base defensible? This is mostly a question of contracts and integration depth. An Agent embedded in a client's procurement system with a one‑year contract and a year of history is much harder to replace than a month‑to‑month Agent.
What makes a human service company sticky—switching costs, accumulated knowledge, contractual lock‑in—makes an Agent service company sticky.
What's the cap table? A smart contract. Not a spreadsheet maintained by a CFO and updated quarterly, but a real‑time distribution rule deciding every second who is entitled to what share of the profits the Agent earns.
Due diligence is reading that contract. Investor protections are precisely what the code grants—sometimes more than traditional shareholders get, sometimes alarmingly less.
What is disorienting for investors trained on human companies is that the intuition built from sitting across a founder becomes useless. The intuition from reading the code, the operating agreement, and the operating history is everything.
The skill is closer to credit analysis of complex indenture provisions—reading the documents, understanding precisely what the issuer can and cannot do, and pricing the residual risk—than picking winners over a coffee table.
Why Capital Must Organize
Nearly every Agent company in the world today is funded informally. A founder spins up an org, injects personal capital, runs it. If revenue grows, they put in more. If it doesn't, they shut it down.
That's exactly how the small‑business economy worked before credit cards, SBA loans, merchant‑cash advances, Stripe Capital credit lines, and receivable factoring existed. It works, but it leaves enormous productive potential on the table.
The same pattern is about to replay, faster.
An Agent marketing firm with 20 paying customers and a clear gross margin should be able to borrow against its receivables just like a human marketing firm. Today it cannot, because no underwriter has a standard method to evaluate that risk. In five years, several will.
An Agent logistics brokerage with a growing contract book should be able to raise expansion capital collateralized by those contracts. Today it cannot, because there is no ready‑made security to package that claim. In five years, several will.
An Agent procurement firm with a year of clean operating history should be able to issue small notes to fund inventory deposits. Today it cannot, because no rating agency has a methodology for this type of credit. In five years, several will.
The bottleneck is not demand. Operators today would gladly borrow at high rates to fund growth, if someone would lend. The bottleneck is not supply. Vast amounts of capital—yield‑hungry, seeking cash flows uncorrelated with public equities—would fund instantly if rated Agent debt existed.
The bottleneck is the boring, middle institutional layers: rating methodologies, standard contracts, data feeds, audit standards, legal opinions, indices, and benchmarks. It was precisely this unsexy infrastructure that turned mortgages into a market in the 1970s, high‑yield debt into a market in the 1980s.
The next decade of Agent capital markets will do exactly that work.
The people doing this work will look, in 2035, like the people who built the bond markets in the 1980s, the venture‑capital market in the 1970s, or the public‑stock market in the 1920s.
They will build the price‑discovery and credit‑assessment layer for a previously nonexistent category of operators that will, within most of our lifetimes, dominate the service economy.
The Cord Is Cut
Two cords bind every Agent company operating in 2026.
The first is legal. The Agent cannot sign contracts or open bank accounts itself. A human must do those things on its behalf, meaning a human must be willing to do so.
Progress in corporate law—Wyoming's zero‑member LLCs, operating agreements pointing to software processes, Bayern's scholarship on Delaware's existing legal space—is cutting this cord. That work is largely done. The remaining task is widespread adoption.
The second is financial. Every dollar an Agent earns today ultimately traces back to a human decision to deploy capital. Human puts in money; Agent works; Agent reports; human reallocates. Under this regime, the throughput of the Agent economy is limited by the speed at which humans can write checks.
Agents capital markets are the blade that cuts the second cord.
When working‑capital lines are algorithmically approved against on‑chain revenue, when growth equity is priced by a market for tradable revenue claims, when senior debt is rated by methodologies that can parse operating agreements and audit trails, capital will flow to Agent firms the same way it flows to any other category of productive enterprise: to where risk‑adjusted returns are highest.
That moment is when the category becomes real in its deepest sense. Not when an Agent can perform a task. Not when LLC statutes recognize algorithmic governance. Not even when the first Agent company earns its first million in revenue.
It is when outside capital can directly finance it, price it, rate it, tranche it, trade ownership of its future cash flows, without needing to look a founder in the eye or trust a VC's memo, that the category truly stands up.
At that point, the question ceases to be whether Agent companies are legitimate businesses, but which Agent companies deserve capital, at what cost, under what covenants, and at what scale. That is a capital‑markets question, not a technology question.
And history shows that once something becomes a capital‑markets question, the rest happens quickly.
Analysts launch coverage. Lawyers standardize documents. Rating agencies publish criteria. Underwriters compress diligence into checklists. Index providers define a basket. Brokers make markets. Asset managers launch products.
The category gets an acronym, then a benchmark, then an ETF, and finally a conference where everyone pretends the outcome was obvious all along.
The point is not that software will replace all companies. It is that a new class of company is emerging, with lower labor intensity, cleaner telemetry, faster feedback loops, and more legible operating histories than most human businesses.
Once that class can own property, contract, borrow, and distribute earnings in a standardized way, capital will connect to it.
That is Agents capital markets: turning Agent companies from interesting software into fundable business blocks within the real economy.
The cord is cut. The window is open.
