These Are a Few of My Favorite Things

After a decade at 500 Global, 50,000+ startups evaluated, 30+ accelerator batches, I have learned to pay attention when the pattern shifts before the narrative catches up. Walking around the Bay this week, the trees are blooming ahead of schedule. The market feels similar.

Within AI, there are two parallel stories that almost never get told in the same room. The loud story is about foundation models and frontier robotics, with hundreds of billions flowing into a small set of foundation labs and embodied AI companies. As of writing in March 2026, OpenAI, Anthropic, and xAI alone account for several hundred billion in cumulative capital raised, with February of this year contributing more than $140 billion in fresh closes between OpenAI and Anthropic.

There is now a real, growing scoreboard tracking AI-native companies that reach $5M, $10M, $100M+ in annual revenue with teams you could fit in a single conference room. Henry Shi's Lean AI Leaderboard. Ben Lang's Tiny Teams. Jeremiah Owyang's research on lean AI anatomy. Spencer Belsky's Lean AI Report. These are not VC marketing projects. They are independent researchers documenting something that keeps showing up in the data no matter how it gets cut.

The old "healthy" benchmark of $200K in revenue per employee now looks quaint. SaaStr recently argued that $500K is the new floor, with the top of the AI-native distribution running multiple times higher.

These are not edge cases anymore. What makes them different is not just that they use AI. It is that they were built around AI from inception. The hiring plan, the org chart, the cost structure, the go-to-market, all designed for a world where intelligence is an API call, not a headcount line item.

The takeaway is not that AI makes companies more efficient. It is that AI-native companies are a different kind of company entirely. The economics only make sense when the whole structure is designed around this premise. You cannot retrofit 2019 SaaS with 2025 AI and expect Midjourney-shaped returns.

When I started asking investors about this category, the same pattern surfaced across different fund types, stages, and geographies. The companies that are working often do not need $20M Series A rounds. They do not want board seats. They have specific asks. The founders are not anti-capital. They are anti-misalignment.

Azeem Azhar at Exponential View has been tracking this from the capital-formation side. His March 2025 essay argued that AI can let founders reach profitability with less capital, and sometimes with very little capital after that. The economics that venture capital was designed for (burn $50M to reach $100M ARR, then raise more to scale the sales machine) do not fit companies that reach $10M ARR profitably with 15 people and keep most of the value for themselves.

I have now heard variations of the same observation from investors in three different Kauffman Fellows cohorts, from former a16z partners who have started micro funds, and from family offices that traditionally allocated to venture but are now writing direct checks to founders. The structural gap is becoming undeniable: the companies that are winning often do not fit the instruments designed to fund them.

Revenue-based financing is coming back, but this time for cash-generative AI companies, not struggling SaaS startups. Pipe's embedded capital product shows the direction of travel: funding businesses through software platforms using predictable revenue rather than a board-seat-and-preference-stack model. The founders get capital without giving up control or upside. It is not classic venture, and it is not a bank loan. It is a third thing.

Hybrid instruments are also emerging. Paddle's AI Launchpad, backed by Founderpath's SaaS fund, is one early example of non-dilutive capital being pointed specifically at AI startups. The important signal is not any single provider. It is that capital is starting to route through revenue and distribution channels, not only priced equity rounds.

Individual investors are also adapting. Dan Rose's recent thread described writing checks to three AI-native companies as revenue shares, not equity investments. "I want to participate in the upside without needing an exit event that may never come," he wrote. "If these companies compound at 40% annually and distribute 60% of profits, that beats most venture returns without the J-curve."

The best companies in this category are largely invisible to traditional deal flow. They do not attend demo days. They do not apply to accelerators. They do not hire investment bankers. They build, they sell, they compound quietly.

They are found in the replies to technical Twitter threads. They surface in the customer case studies of infrastructure companies like Stripe and Anthropic. They appear in the "built with" sections of products you already use. They are everywhere except demo days and pitch competitions.

Indie Hackers' AI-native leaderboard now tracks 847 companies with verified monthly recurring revenue above $10K. Most are bootstrapped. Most have fewer than 10 employees. Most would be venture-scale opportunities if they wanted to be, but the founders have chosen a different path.

These predictions matter because they represent different possible futures for how capital and company-building intersect. If they prove accurate, the venture capital model as we know it becomes one option among several, not the only path to building valuable companies at scale.

What I find most compelling about this moment is not that AI makes things more efficient. It is that AI-native economics create space for different values to reassert themselves. The best founders in this category care more about building something that lasts than building something that exits. They want to be compensated from the value they create, not from the value they might create if someone else buys them later.

This is not anti-growth. This is not lifestyle business thinking. This is a different theory of how valuable companies compound and who should benefit from that compounding. The founders building these companies often have more ambitious long-term visions than their venture-backed counterparts. They just want to build them differently.

Over the next twelve months, watch for several inflection points:

Fund launches with explicit AI-native positioning. The marketing will be "we back companies that do not need $20M Series A rounds." The thesis will center on revenue efficiency, not growth at all costs. The check sizes will be $250K-$1M, not $5M-$15M.

Major secondary platforms adding AI-native company coverage. Forge, EquityZen, and similar platforms currently focus on late-stage unicorns. They will add earlier-stage, profitable AI companies as a distinct asset class. This creates liquidity for employees without exit pressure on founders.

Revenue-based financing scaling beyond individual experiments. Currently, RBF for AI companies happens deal-by-deal through individual funds or family offices. Institutional products with standardized terms and systematic origination will emerge.

Geographic diversification of the pattern. The examples I have cited here are mostly US-based. The economic logic works globally. Watch for similar companies emerging from other ecosystems where venture capital penetration is lower but technical talent density is high.

Even if you are not building an AI-native company, these dynamics will reshape the broader startup ecosystem. When the most capital-efficient companies can access non-dilutive funding, traditional venture-backed startups will face different competition for talent, customers, and market position.

For investors, this means that deal flow increasingly splits into two distinct categories: companies that need significant upfront capital to prove product-market fit, and companies that have already achieved sustainable unit economics and want growth capital without control dilution. The second category will require different fund structures, different LP relationships, and different success metrics.

For employees considering whether to join an early-stage company, the equity value proposition becomes more complex. A traditional startup equity package is a bet on a future exit event. An AI-native company with profit sharing is current income plus ownership in a sustainable business. The risk-reward calculation changes fundamentally.

I believe we are watching the emergence of a parallel track for company building that will prove as significant as the original venture capital model was when it first developed. The difference is that this track leads to permanent, profitable, founder-controlled companies rather than temporary growth vehicles optimized for acquisition or public offering.

Whether this becomes the dominant model depends on whether the infrastructure develops to support it at scale. The early signals are promising, but infrastructure changes slowly. The companies themselves are not waiting. They are building the future they want to work in, one small, high-leverage team at a time.