Author’s Note: This is a critique of funding structures and incentives—not of the people or projects working within them. Plenty of sponsorship, patronage, and acquisition models do great work. I’m focused here on durability, and on the tradeoffs that matter for how I want to build.

There’s a category of “north star” companies: small, tight-knit teams building impactful open source projects, funding slow growth and focused scope with commercial products and services that support an open ecosystem.

These companies emerged from the early, experimental internet. Teams that cared about craft and built serious, long-lived projects without becoming hobbies or startups. That path once made sustained, independent work possible. Today, it no longer seems viable by default.

The minimum viable credibility required to be taken seriously is time-intensive—and therefore expensive.

In the wake of widespread dependency vulnerabilities, proactive maintenance and strong governance are no longer optional. As humans—and increasingly agents—interact with our systems, documentation becomes critical infrastructure. Human-translated docs open projects to a global pool of contributors. Thriving ecosystems rely on healthy, diverse communities; without them, we end up dependent on actors whose incentives are often misaligned with the public good.

AI is accelerating a trend that began years ago. As the hypergrowth, unicorn era took hold, the narrow path to viability for middle-scale projects was squeezed shut. The investments required to compete rose alongside expected outcomes. Even with AI-assisted development, the cost of entry for serious, production-critical open source projects is easily measured in millions of dollars of combined time and labor.

Legitimacy is expensive. The costs are front-loaded.

This isn’t new. Open source has always struggled with funding, coordination, and burnout. What is new is the widening gap between productivity and trust. AI compresses output, not legitimacy: it can help write code, but it doesn’t maintain it, govern it, document it, translate it, or earn the confidence of the communities and companies that rely on it. Meanwhile, post-ZIRP capital tightened just as expectations around security, compliance, and reliability increased. The result is a larger, more expensive credibility gap—one that disproportionately punishes small, serious teams before they’ve had a chance to prove themselves.

For corporate sponsorship to work, projects must already have adoption and scale. Community funding without that foundation requires a contributor pedigree and mutual-aid infrastructure that are still emerging. Patronage—maintainers employed through goodwill or as part of cloud-service funnels—was often contingent even at its peak. When belts tightened after ZIRP (circa 2022), these projects were reclassified as discretionary and reprioritized.

Funding options while the risk is highest are few and far between.

In that environment, it’s easy to see why ambitious engineers turn to venture capital, often paired with a future cloud offering to justify it. I’ve seen how effective that can be. Venture funding buys time: hiring contributors, investing in documentation, supporting a community, and building an ecosystem that can compete with incumbents. But that freedom comes at a cost.

Incentive alignment alone rarely survives exponential growth expectations. Some projects can carve out sustainable niches, but most open source efforts will never reach a market large enough to satisfy those demands. When growth becomes the dominant constraint, stewardship is often reframed around it.

I respect founders whose ambition requires venture-scale investment—but the set of acceptable outcomes is regrettably narrow. Many of the most valuable open source projects are incompatible with those incentives over the long term.

The early internet was a place of weird art and passion projects. The middle path made that viable. It was filled with explorers experimenting with different ways to sustain their work. The groups that endured tended to be democratic, open, and unconvinced by the allure of hypergrowth.

So where does that leave us?

Thankless maintenance leads to burnout. Sabbaticals rarely sustain solo maintainers. Niche projects struggle against VC-backed hyperscalers. Paid services are hollowed out by AI. Agencies split focus. Corporate sponsorship can drift toward anti-competitive pressure.

These models share a failure mode: they externalize risk onto maintainers while demanding increasing focus drift.

Venture capital, as practiced today, tracks Big Tech booms and busts. Platform dominance and functional monopoly are the endgame. That alignment is geographically and culturally narrow, rooted in U.S. financial systems, at odds with open source as a global commons. Community and stewardship can exist within that model, but they are means to an end—not the end itself.

If the missing middle deserves to exist again, we have to build it.

That means investing in art, diversity, and creativity. It means addressing the internet’s original sin by building protocol-first commerce¹. It means funding shared safety nets and taking user empowerment seriously, taking a cue from the atproto ecosystem and building protocols before products. Over time, we’ll need ways to pool risk across maintainer-driven projects without forcing each one to become its own startup. Goodwill alone isn’t enough. Hope alone doesn’t scale.

When empires refuse to provide the foundations for us to succeed, the internet lets us build them ourselves. Same as it ever was.

As I start to invest my focus in two long-term projects—Bombshell and LGTM Supply Co.—I’m choosing to face these constraints head-on. I’m not interested in funding models that repeatedly fail promising work. I want to build slowly, sustainably, and scale appropriately.

I want to live in a world where efforts like these can thrive. Watching projects in the atproto ecosystem shift our focus back to collaborative, open protocols feels like a glimpse into that future. Supporting the full-time focus of the people doing the work should be the lowest bar to clear.

Footnotes

1. regarding crypto bros—you do not, under any circumstances, “gotta hand it to them” ↩