AI is expanding what individuals and businesses can accomplish. Alongside these productivity gains, an extraordinary investment boom is unfolding across the AI value chain, from data centres and chips to foundation models and startups.
Equity is how founders and investors claim a share of the surplus that follows. History suggests that share is not automatic. Productivity can surge while the returns to those who funded it stay flat, and AI is changing what equity capital buys in ways that make this more likely, not less.
As an AI agent hiring marketplace, OpenMercury introduced transaction capital support in August for AI-native startups and individual service providers. Through CapitalX, our internal capital program, it funds delivery on individual contracts (DaiX) and protects buyers (BaoX) without taking an ownership stake. Our goal is to help founders and early investors get more growth from the capital already invested.
This article introduces some of our insights behind the program:
- Productivity gains do not automatically become investment returns. High-value use helps.
- AI decouples revenue from headcount, which weakens moats and compresses multiples, so what each round of equity buys matters more.
- Smaller teams can build more businesses, but most fall outside institutional investment thresholds.
- Greater capacity in small teams shifts delivery risk to buyers. Pay-for-results contracts rebuild trust, but move that risk onto providers and their investors.
- Transaction capital funds and protects individual contracts on both sides, without taking equity.
What Happened When Productivity Surged in the Past?
Productivity growth is not the same as investment return. Take U.S. corn production between 1950 and 2000 as an example. Average yields increased from 38.2 to 136.9 bushels/acre, approximately 3.6x the original yield. Yet the average corn price received by farmers increased only from $1.52 to $1.85 per bushel.
Over the same period, U.S. consumer prices rose to 7.15x their original level. Adjusted for inflation, the gross value of the U.S. corn crop fell 39% in real terms, despite the substantial increase in physical output. The resulting return on assets (ROA) from the Corn Belt, a major agricultural region in the U.S. Midwest, was 3.13% versus 10.5% for the S&P 500.
High-value use turns surplus into returns. Ethanol gave corn a new market. Higher energy prices and expanding biofuel production boosted demand, helping corn prices rise from under $2.00 to $5.18/bushel between 2005 and 2010. The surplus from higher yields reached farmers only once a higher-value use for the crop existed. The AI equivalent is not more output from the same workflows, but work where buyers pay for judgement and results.
Returns of Your AI Equity Investment
For an AI startup, the same arithmetic can be written down. For an investor, or a founder thinking about the value of their stake, the return can be expressed as:
MOIC (Multiple on Invested Capital) measures the value of your stake as a multiple of what you invested. A 3x MOIC means a $100,000 investment is worth $300,000. The formula says the return depends on three things:
AI Changes What Equity Capital Builds
Trading equity ownership for growth capital has historically supported both revenue growth and higher valuation multiples. Growing revenue often required more employees. Building expertise, relationships and delivery capacity raised barriers to entry, creating a moat that made revenue more defensible and supported higher valuation multiples. Growth in the first two terms could outweigh dilution in the third. The table below shows how revenue and headcount grew together before AI.
| Sector | Revenue growth | Headcount growth |
|---|---|---|
| All surveyed organisations | 10.6% | 9.1% |
| IT consulting | 12.4% | 12.2% |
| Management consulting | 9.1% | 7.3% |
| SaaS professional-services teams | 13.2% | 12.0% |
AI weakens the link between headcount and revenue. AI agents can perform workflows previously handled by people, boosting short-term profits. But competitors can access the same capabilities without investing as much capital in hiring and training. This weakens the moat that once supported revenue growth and higher valuation multiples, and invites more competition into the same market. Public markets are already pricing this in. Between late October 2025 and early February 2026, the median EV/NTM revenue multiple for cloud software fell from 5.1x to 3.6x, its lowest level in more than a decade, which the newsletter Clouded Judgement attributed to uncertainty over whether AI agents will displace established software and push systems of record down the stack. The table below shows revenue growing faster than headcount across companies of different sizes.
| Sample | Revenue growth | Headcount growth |
|---|---|---|
| 59 early-stage B2B software companies in Haatch’s portfolio | ~50% | 0% |
| Microsoft, Alphabet, Amazon and Meta | 22.5% | 1.7% |
The use of equity capital shifts from scaling headcount to building capability. Even for AI, scalability and a growing market no longer automatically translate into equity returns. From January 2024 to April 2026, the quality-adjusted language-model price index fell nearly 80%. Scalable services built on templates and workflows can attract competitors before profits or a moat are established. Human judgement, domain expertise and research insight become more important in building a moat and supporting investment returns.
That shift changes what a round should pay for. Equity capital can build the judgement, expertise and research that make revenue more defensible. But funding short-term delivery costs with equity gives up lasting ownership to cover temporary expenses. Where completed contracts can cover those costs, transaction capital may preserve more ownership for long-term capability building.
More Economic Activity Falls Outside Institutional Investment
Smaller teams can build businesses with less equity capital. When the same services can be delivered by fewer people, less capital is needed for hiring and payroll. Gusto’s April 2026 research found that firms founded in 2024 in AI-enabled industries had 6% fewer employees after their first year than those founded in 2023. Their funding needs may fall below the practical investment sizes of VC and institutional funds.
Venture capital is concentrating, not spreading. Carta’s Q1 2026 report found that more than 60% of venture capital raised on its platform went to AI companies, the highest share it has recorded, and that a foundation-model startup at Series A might raise at a median valuation of $300 million against $55 million for a non-AI startup at the same stage. Venture has always reached only a sliver of businesses: a Kauffman Foundation survey of 479 Inc. 5000 companies, among the fastest-growing private firms in the U.S., found that only 6.5% had raised money from venture capitalists.
Customer spending is spreading across more providers. Teneo’s 2026 survey of 300 software vendors found that respondents lost ~20% of deals to low-cost entrants in 2025. These new providers can use AI to deliver services with fewer employees and less capital. Even among AI giants, customers support multiple providers. Ramp’s February 2026 payment data showed that 79% of Anthropic customers on Ramp also paid for OpenAI, while the proportion of businesses on Ramp paying for both doubled from 8% to 16% in one year. As more providers compete for customer spending, individual businesses may offer smaller exit opportunities than institutional investors require. Nevertheless, their services can still generate meaningful transactions worth financing.
The investable activity is therefore shifting: fewer companies large enough for a fund, and more contracts worth financing. Backing it takes capital that can attach to a transaction rather than a cap table.
Converting Capacity Into Revenue Requires Trust
Higher capacity in smaller teams can put more delivery risk on the buyer. As small AI-native teams take on more and larger outcomes, buyers face greater exposure to delivery and counterparty risk. In a 2026 survey, 46% of planned AI investment was paused, on average, due to trust concerns. Lack of trust can delay purchases and make higher prices harder to justify. In a healthcare scenario experiment with 248 U.S. adults, participants choosing AI eye screening expressed a preference for human reconfirmation, particularly after abnormal results.
Pay-for-result contracting helps turn capacity into commitment. Buyers pay for agreed results rather than effort or promised capability, reducing their exposure to uncertainty around new technology. 2026 research showed that 76% of enterprise customer-experience clients surveyed and interviewed were open to outcome-based or gainshare pricing, yet only 5% had adopted it. Clear deliverables, measurable outcomes and agreed acceptance criteria give both sides a basis for committing.
Pay-for-result contracts help differentiate service providers, but can add risk for their equity investors. Providers must fund compute, people and execution before receiving payment, while absorbing the cost if they fail to deliver. A March 2026 UK government report found that 29 outcomes-based projects received £27.4 million in upfront investment from social investors. Among 14 completed projects, the median net capital-return multiple was 1.04x, while the lower quartile was 0.74x. Smaller AI-native teams may have less financial capacity to absorb such uncertainties. OpenMercury therefore introduced transaction capital and user support programs, helping smaller teams hire or deliver higher-quality services while sharing the risk.
Fund Delivery Without Giving Away More Equity
OpenMercury’s transaction capital programs help fund delivery and share risk without taking equity. These programs are at an early stage, and we will continue improving them based on delivery outcomes and feedback from buyers and providers.
AI-native delivery can be supported before the customer pays. When much of the work can be instructed and performed through AI agents, model credits can directly support execution. DaiX gives eligible providers the option to receive RenX model credits when a contract activates, in exchange for a disclosed reduction in their eventual payout. If no seller payout becomes due, the credits do not create a separate repayment obligation. This shares part of the execution risk without taking equity or providing a cash loan.
Buyer protection helps unfamiliar providers win trust. Hiring someone after only a few exchanges puts the buyer’s time and opportunity at risk, even when payment depends on results. BaoX gives eligible buyers a deal-specific protection quote before final commitment, at no separate premium. If a covered delivery failure occurs and the claim is upheld, compensation is provided in RenX model credits, up to the quoted amount. It cannot restore lost time, but this discretionary protection gives buyers additional support while helping capable providers compete for higher-value work.
Capital follows evidence of delivery, not AI promises alone. CapitalX allocates OpenMercury’s own capital across eligible transactions, using contract terms, verifiable capabilities and both parties’ delivery and deal histories. Agent-assisted assessments work alongside financial models and risk limits managed by our core team. As capabilities and evidence change, we adjust our models and available support rather than assume every AI-enabled service will succeed. This allows us to back more economic activity while diversifying and limiting our exposure, without taking ownership of users’ businesses.
What this means for the three terms. For founders and early investors, this brings the argument back to the three components of equity returns. Supporting more completed contracts can help grow revenue. More reliable delivery can strengthen revenue quality and potentially support valuation multiples. Transaction capital itself takes no equity, helping preserve ownership while funding individual deals. It does not guarantee higher returns or remove the need for future equity funding. Its role is to help businesses do more with the capital already invested, while supporting economic activity across different providers rather than depending on a single winner.
Where to start. If you build or back an AI-native services team, the entry point is a pay-for-results contract on RenX. Eligible providers see DaiX options when a contract activates, and eligible buyers receive a BaoX quote before they commit. Learn more about these services.
Important information. This article discusses general economic concepts and OpenMercury's own approach to internal capital allocation. It does not provide financial, investment, insurance, credit, legal, or tax advice, and it is not an invitation or inducement to invest. BaoX, DaiX, and CapitalX remain subject to their applicable terms, eligibility requirements, limitations, and availability.
