Part four · First principles ·
The Collateral Limit.
Firms will not just survive the age of super intelligence. They will thrive, because they can be insured. A firm of one can compete on one condition: someone backs it up to a limit set by what it has posted, not by risk nobody can model.
By John Rector · about 9 minutes
The jobs your limit can reach
A firm of one, backed. Her collateral sits in trust, and a backer writes a limit against it. Every job inside the dome is one she can now sign for.
- The limit someone will stand behind
- A job within reach: it can be contracted
- A job out of reach: it may as well not exist
When you cannot measure the risk, measure the backing.
Why firms thrive
A firm can carry a limit. It has a balance sheet, a loss history and a policy big enough for the customer’s risk. Customers buy that limit as much as they buy the product.
How a firm of one competes
Not by being liable alone. By being backed: collateral in escrow, and a new kind of backer willing to write a limit against what she has posted.
Wednesday, 10:20 a.m.A hospital’s purchasing manager likes the scheduling system a firm of one built better than the big vendor’s. Then the contract asks for a certificate of insurance with a limit larger than her whole business is worth. The big vendor sends theirs within the hour. She has nothing to send. The better product loses.
What firms are for now
In the first part of this series I argued that super intelligence makes hands nearly free, so the old reason for firms, the cost of organizing people, is dissolving. In the second I argued that the one thing a delegate can never be is liable, so a human has to stand behind every piece of work. In the third, that the best-paid human work until 2030 is the work an insurer will cover.
Put those together and something surprising falls out. Firms will not fade. They will thrive, for a reason Coase never needed: they can be insured. A firm has a balance sheet, a record of claims, and a policy whose limit is big enough for the customer’s risk. When a hospital, a city or a bank buys from a firm, it is buying the product and the limit together. Often the limit decides.
Liable is not enough
Liability is only worth what you can pay. A firm of one is liable down to her last dollar and not one dollar further. The customer’s real question is never just who is liable. It is: up to how much?
That is why the scene above ends badly even though everyone in it acted sensibly. The purchasing manager is not choosing the bigger firm because she trusts its people more. She is choosing the bigger limit, because her own job depends on someone being able to pay if this goes wrong.
The customer’s question is never only who is liable. It is: up to how much?
People have always pooled to reach a bigger limit
Lloyd’s of London solved this problem three centuries ago with syndicates. No single Name could stand behind a whole fleet, so Names joined syndicates, and each member was liable only for a pre-agreed percentage of the syndicate’s business. Together they could carry risks none of them could carry alone. A firm is a syndicate in disguise: a pool of capital that lets a group of people promise more than any one of them could.
A firm of one has no syndicate. In 2026, that is the real wall, not talent and not tools.
The escrow, and the backer that grows out of reinsurance
The simplest backing is an escrow: money set aside with a third party, released when the job is done. Construction already runs a more elaborate version. Before a surety issues a bond, the contractor signs a general indemnity agreement, and the owners usually agree to indemnify the surety personally, putting their own assets at risk. Sureties may also ask for cash or a letter of credit. People in the industry describe surety as a credit product, closer to a loan than to a policy. It is underwritten on what the contractor can put behind the promise.
At the very top of the insurance market there is a structure built on exactly this principle. In collateralized reinsurance, the backer funds a trust account equal to the limit of the coverage it writes, and the money comes back if there are no losses. Because the cover is fully collateralized, it lets unrated outside capital, from pension funds to hedge funds, take part in reinsurance. The limit is simply the collateral.
Point that structure down at people and you have what the firm of one needs. A backer, an offshoot of reinsurance, holds an operator’s collateral in trust and writes a limit against it. It underwrites the collateral more than the risk. That is not a weakness. It is the whole point. Insurers are carving AI out of their policies because they say they cannot model it. Collateral needs no model. You can count it.
Who supplies the collateral can vary: the operator’s savings, a paid-off truck, a share of future receivables, a family member who believes in her, investors who want the return. My argument, not yet a record: whoever builds this backer at scale will do for firms of one what the corporation did for the merchant.
Wednesday, 10:20 a.m., 2029The same hospital, the same purchasing manager, a different firm of one. Her proposal carries a verified limit: collateral in trust, a backer’s name, and a number the hospital’s own AI checked before the meeting. The contract is signed by lunch.
Five predictions
Firms thrive, redefined as insurance pools.
Their edge will be the size of the limit they can carry, not the number of people they employ. Some of the most valuable firms of the 2030s will be small teams standing on very large limits.
A new kind of backer appears between escrow and reinsurance.
It will write limits for firms of one priced mostly on collateral posted, not on modeled risk, and it will be built first for exactly the AI-delivered work that standard carriers now exclude.
Backed limits become public and machine-readable.
A customer’s AI will check an operator’s verified limit before it ever shortlists her, the way it checks a license today.
Firms of one form syndicates again.
Operators will pool collateral to reach limits none could reach alone. It will be the Names model with one difference: each member’s exposure stops at what she posted.
Collateral becomes the solopreneur’s scarcest asset.
Keeping expenses low used to buy freedom. Now it also builds collateral, and collateral is what buys the right to bid on big work.
What this changes in my own advice
For years I have told my students to let their first employer be their first customer, and that this only works if they keep their financial needs low: rent rather than buy, ride rather than own. I used to say that buys freedom. It also builds collateral. Every month of low burn is a deposit toward the limit that will one day let them sign for work bigger than themselves. At our incubator, the founders I meet can already build a product and reach real customers on a budget. The wall I expect them to hit next is the certificate of insurance. The ones who planned for it will walk through.
The claim, and how it fails
- Already true
- Collateralized reinsurance funds a trust equal to the limit. Sureties underwrite on indemnity and collateral, often the owners’ personal assets. Lloyd’s syndicates pooled individuals by pre-agreed shares. Carriers are excluding AI from standard policies.
- What has to happen
- Customers and regulators accept a verified, collateral-backed limit in place of a traditional certificate of insurance. Holding collateral in trust becomes cheap and fast enough for a one-person business.
- Where I am probably wrong
- If the big carriers learn to price AI risk by 2028, ordinary insurance comes back and the collateral backer stays a niche. Worse, if the collateral required is so high that only the already wealthy can post it, the age of super intelligence will favor capital over ideas. I would count that as a failure, not just of this prediction but of the opportunity.
The idea found its person. The person found her hands. Now she needs a backer.
Background
- Guy Carpenter, Features of Collateral Structures (2015).
- Artemis, What is collateralized reinsurance?
- Axcess Surety, Surety Bonds and Indemnity.
- Marsh McLennan Agency, General Indemnity Agreements.
- Ronald H. Coase, “The Nature of the Firm,” Economica 4 (1937).
- Part one: We Don’t Generate Ideas, part two: Liable, and part three: Paid to Be Liable.