Code is cheap, and the market has noticed
For forty years software was the scarce thing. A company that owned good code owned a moat. In April 2026 Sundar Pichai said 75 percent of all new code at Google is generated by AI and approved by engineers, up from 50 percent the autumn before.
The market repriced. On February 3, 2026 about $300 billion of market value left software and data companies in a single session, after one AI product release.
Releases from ten major labs, counted on September 1, 2026. A plan scoped today is built on last quarter’s models.
Sources
Sundar Pichai, Google Cloud Next (April 22, 2026) · Forbes, $300 Billion Evaporated (February 4, 2026). The $300 billion is Forbes’s estimate across a basket of software and data companies · Keenable, from the Top Frontier LLMs Release Tracker
The value moved to the work
Software was always the smaller number. For every dollar spent on software, six are spent on services.
Most of that spend is payroll. It is the people carrying decisions between systems inside insurers, banks, brokers and accounting firms. It is not a market anyone sells into. You have to be inside it.
That work sits inside businesses that already run. They hold the customers, the licences and the people who know the exceptions. Code cannot write one of those.
Sources
Sequoia Capital, Services: The New Software (March 5, 2026) · General Catalyst, The Future of Services (August 28, 2025), which puts US service industries at more than $6 trillion a year against about $370 billion for software
Buying the tool is not rebuilding the business
Most businesses have bought AI. Few have put it to work.
Two different measures of US businesses. Paying: Ramp, October 2026. Using in any function in the past two weeks: US Census Bureau, May 2026.
The spend gap
The median buys seats for its people. The top one percent pays engineers to rebuild the work itself. Ramp AI Index, August 2026 data.
The talent gap
One person in four hundred has seen what these systems can do. The businesses that need to rebuild do not employ them.
Have barely touched it.
Use a free chatbot.
The people a rebuild needsWatch weeks of work collapse into a prompt.
Have access to frontier systems internally.
Estimates by Andrej Karpathy, October 11, 2026. Band widths on a log scale. This talent does not take a job in an established business. It comes in as an owner.
Why the old routes do not close it
| Route | Its clock | What that means |
|---|---|---|
| Consultants | By the week | Paid by the man-week, in scopes sold into larger scopes. A longer project pays more |
| Software vendors | By the seat | Sell a tool by the seat, then wait for someone inside the business to make it work |
| Private equity | By the fund | Change bought from consultants and vendors, with an exit date set before the work starts |
AI’s clock is 4.3 days. The capability has to live inside the business.
Sources
Ramp AI Index · Ramp AI Index, September 2026 · US Census Bureau, Business Trends and Outlook Survey (May 26, 2026) · Andrej Karpathy (October 11, 2026)
Build a new one, or buy the one that runs
With code cheap, capital is making two bets on how the gap closes.
New companies that do the work themselves
- Licences, regulators and capital requirements to win first
- Every customer still to earn
- Years before the moat exists
- Capital priced as venture risk
Buy the business and rebuild it
- Customers, licences and a balance sheet on day one
- The moat already built
- Uplift measured against a business that already runs
- Capital priced against a working asset
The second bet in 2026
The Baldwin and ADT transactions are expected to close in the first quarter of 2027.
Bet one: who has been funded
Capital and traction are company-reported. We found no measured loss of market share at a named incumbent in 2025 or 2026, and no newly licensed AI-native bank or credit union.
Bet two: who is buying what
Figures are company-reported or reported by the lead investor.
The owners who keep what they hold
A third group is neither building nor buying. Private equity firms and pension funds already own thousands of running businesses, and they are paying to bring the capability in.
Each commitment secures the same thing: people who can go inside a business and change how it works. The delivery companies earn fees. The owner keeps the gain.
The commitment register
What the results show
A few of the buyers now publish what happened after they went in.
Reported by the companies or their investors. Not independently audited.
The gain arrives as capacity: the same people handle more work. The established relationships stay. And in every case engineers worked inside the business alongside the people who run it.
What we take from all of this
The work is the prize, and it cannot be sold into.It is payroll inside businesses that already run. The only way to reach it is to be inside the business.
The moat is whatever code cannot write.A licence, a customer book and a balance sheet take years to build. Where they do, buying into the business that holds them beats starting a new one.
The constraint is people.One person in four hundred has seen what these systems can do, and they do not take jobs in established businesses. They come in as owners.
The capability has to live inside the business.The tools change every few days. A plan handed over by an outside firm is out of date before it is finished.
The gain belongs to whoever owns the business.Every result on record came from a buyer that owned what it rebuilt. Whoever does the rebuilding should hold a piece of it.
We rebuild businesses on AI. Then we own a piece of them.
Zeroe is in the second bet. We buy into running businesses with room to grow. Four things have to be true: a moat, underused capacity, aligned ownership and a large addressable market.
Why this team
A rebuild needs two kinds of people: those who see what AI can do, and those who know how to change a business. Zeroe is both, in one team, inside the business.
| Name | Before Zeroe | At Zeroe |
|---|---|---|
| Farouk Jivani Co-founder and CEO | More than 20 years of CEO-mandated turnarounds across banks, telcos, hospitals, ports, energy and mining | Leads the deals and the blueprint |
| Ali Najafian Co-founder and CTO | More than 25 years shipping platforms. Founder with three exits | Leads the rebuild and the agent systems |
| Eric Espinosa Head of Strategic Partnerships | More than 20 years building alliances across municipalities, the UN and major brands | Builds the network the deals come through |
| Julien Cleret Head of Deployment | More than 20 years rebuilding businesses by hand, in consulting, at Ingersoll Rand and at Honest Company | Leads the team inside each business |