I run a small open source web operations company in Seattle. We've been in business since 2003 -- the same year I wrote Open Source Solutions for Small Business Problems, a book arguing that small businesses should own their software systems rather than rent them from platforms they don't control. Twenty-three years later, that argument has only gotten more urgent. I want to say plainly what we're about now, because I suspect that saying it plainly is itself a useful thing to do.
Freelock bets on diversity, openness, ownership, and human-scale systems. We bet against consolidation, vendor lock-in, and the prevailing story that says enormous platforms run by enormous companies are the natural and inevitable shape of the future.
We are not neutral about this. We think consolidation is harming people. We think inevitability is a marketing campaign. We think the future of the internet is still genuinely up for grabs -- that it can still be open, federated, diverse, and oriented toward human needs rather than enclosed, monopolized, extractive, and oriented toward shareholder returns -- and we're placing our work on the side of the people who want to keep it that way.
Inevitability is a sales pitch.
This is not a popular position to take loudly in 2026. Saying so anyway is part of the point.
What we're against
Three things, mostly.
Industry consolidation
The steady process by which a handful of enormous companies acquire their competitors, enclose what used to be open standards, and convert shared infrastructure into platforms they rent back to the rest of us.
Salesforce has done this with Heroku, Slack, Tableau, MuleSoft, and a dozen others -- ending Heroku's free tier and stranding an entire generation of developers who had learned the trade there. Adobe tried to do the same to Figma; the regulators stopped them, but the design community remains exposed. Microsoft has done it across an empire that now includes GitHub, LinkedIn, a growing chunk of the AI ecosystem, and the near-monoculture of corporate offices on Microsoft 365 and Teams. Google has done parallel work with Workspace and Drive, and has shut down so many products its users depended on that a community-maintained graveyard now tracks them.
Cloudflare doesn't acquire companies the same way. It has just become so essential that nearly half of the most-visited websites in the world route through it. When it had a bad day last November, streaming services and AI platforms and communications tools went dark together -- because everyone had quietly become dependent on the same single point of failure.
A monoculture waiting for its blight.
The narrative that AI is inevitable -- and that the version we're being given is the only version possible
A handful of enormous companies are racing to build the largest possible general-purpose AI models, drawing extraordinary quantities of capital, water, and electricity from communities that did not necessarily agree to provide them, trained on the work of humans who were not asked and are not being compensated.
We use AI extensively at Freelock -- you can read our AI use policy for the specifics -- because the leverage it provides is real, and refusing to use it while competitors do is not a serious option.
But the AI we want is not the AI we're being sold. We want smaller, targeted, often open source models that do specific things well, that can be inspected and modified, that run on reasonable hardware, that don't require a billionaire's data center to operate. We want a future in which the technology is plural rather than monopolized, serving human work rather than replacing the humans who do it.
That future already exists, in glimpses. In March 2026, a VFX artist named Niko Pueringer released CorridorKey -- an open source neural network that solves a thirty-year-old problem in green screen keying, runs on a consumer GPU with 6-8 GB of VRAM, and was built by someone who actually does the work it automates. It got 8,400 GitHub stars in its first couple weeks and was integrated into other tools within a month. It is exactly the kind of AI the big AI companies aren't building: domain-specific, transparent, modest in its resource requirements, made by a practitioner for practitioners. This is the AI future that's actually showing up. It just isn't the one being marketed.
The version of AI being marketed as inevitable is none of these things. And the public is starting to notice -- Adam Conover recently walked through the polling data: bipartisan opposition to data centers, Gen Z approval of AI in free fall, even Pope Leo dedicating a major encyclical, Magnifica Humanitas, to safeguarding human dignity in the time of artificial intelligence. The pushback is bigger than Sam Altman and Marc Andreessen would have you believe.
The billionaire class as a sign that something is structurally broken
A society that produces individuals with more wealth than entire nations, while ordinary people cannot afford housing or healthcare, has failed at the basic task of distributing the gains of collective effort.
The tech industry has become particularly efficient at producing billionaires, and the technologies it builds increasingly reflect the priorities of the people funding them. Venture capital is in the business of finding the next monopoly, not the next sustainable cooperative. The current shape of the internet is the shape that maximizes billionaire production, and it has been shaped that way on purpose.
This pattern isn't unique to tech. It's everywhere you look. Hospital systems acquire independent practices. Private equity buys out veterinary clinics and dental offices and family newspapers and funeral homes, extracts what it can, and moves on. Grocery chains consolidate. Local radio stations get absorbed into national networks that broadcast the same content in every market. A friend of mine recently retired from a business he founded and ran for thirty years; he sold it to a private equity firm, stayed on as CEO for another year, and then walked away. For him, individually, it was the best deal available -- a fair return for thirty years of work, and I can't fault him for taking it. But add up a million decisions like that one and you get enclosure by another name: a country with fewer owners, fewer independent businesses, fewer places where local people answer to local people. The historical commons was enclosed by acts of Parliament and barbed wire. The modern commons is enclosed by acquisition. The accumulation is the problem, even when no single decision is.
Tech is where the extraction is most extreme, and most visible. But it is not where the extraction is unique.
The historical commons was enclosed by acts of Parliament and barbed wire. The modern commons is enclosed by acquisition.
What we're for
Free and open source software, as a political achievement worth defending
In 2009 I flew to Helsinki to spend a weekend with Monty Widenius, the creator of MySQL, asking how a small open source business could survive. Over a long dinner he explained why he had chosen the GPL for MySQL. It wasn't ideology. It was defense. The copyleft requirement -- that derivative works also be GPL -- meant no competitor could take his work, fork it into a proprietary product, and use it against him. The license was the mechanism that protected the project from capture, while still letting him build a business on top of it. If he were starting today, he told me, he'd choose the AGPL, which closes the modern SaaS loophole. The license was the moat.
Sixteen years later, Monty's current project, MariaDB, moved to the Business Source License -- which keeps a similar defensive posture but pushes harder, reserving commercial control to the company for several years before the code reverts to a fully open license. I don't claim to know what conversations led there, and I'm not in a position to judge from outside. But the trajectory matters. Even the founders who built open source with the clearest defensive intent have ended up under pressure to close it up further. The pressure is structural.
The pragmatic "open source wins because it's everywhere" framing has been the dominant story in tech for twenty years. Open source has indeed won enormous market share. But the permissive licenses (MIT, Apache, BSD) favored by that framing are exactly the licenses that allow enclosure to happen. The copyleft instruments of the free software tradition were always the stronger defense, and they have been quietly out of fashion in VC-funded open source for over a decade. The capture we're now living through is the predictable consequence.
When the structural pressure does hit a project we depend on, there is another community defense besides the license: the fork. When the developers or the community lose confidence in the trademark holder, they can build a parallel project on the open foundation and follow the developers who built it. We have followed community forks consistently. We moved from Mambo to Joomla in 2005, from SQL-Ledger to LedgerSMB in 2006, from ownCloud to Nextcloud when Frank Karlitschek led the same fork move out of the company he had founded, and more recently from Gitea to Forgejo when Gitea's maintainers transferred the project to a for-profit company. We are migrating from HashiCorp Vault to OpenBao right now, following the community fork that emerged after HashiCorp moved its products to the Business Source License in 2023 -- the same licensing move that prompted the OpenTofu fork of Terraform. HashiCorp itself was acquired by IBM in early 2025. The pattern arrives on schedule.
The pattern extends well beyond HashiCorp. Redis, MongoDB, Elastic, and MariaDB have all made similar moves in recent years. The shape is consistent: VC-funded startups build on the open commons, achieve scale, get pressured by investors to return profit, and quietly close up the very openness that allowed them to grow.
The fork is the community's safety valve. The license is what makes the fork possible.
We come down hard on the community side. Software shared freely as a commons produces more durable, more secure, and more adaptable systems than the proprietary alternative. That this works at all flips a lot of capitalism's assumptions on their head, and we should be much more loudly proud of it than we are.
Owning the stack, renting the metal
We don't run servers in a closet. (Well - we have a couple...) We use cloud infrastructure -- currently AWS, Google Cloud, DigitalOcean, and Linode -- spread across providers so no single one can hold us hostage. If any of them gets hostile, raises prices unreasonably, or has a bad enough day, we can pull up stakes and move. The cloud is plumbing. The plumbing is fine.
What we won't do is rent the house from the plumber.
The layer above the metal -- the operating system, the applications, the data, the configuration, the institutional knowledge of how it all works together -- that we own. Almost all of it is open source. The exceptions are deliberate and minimal: a few specialized tools where the proprietary option still does something open alternatives don't, network gear from Unifi, developer tools like PhpStorm and Claude Code, an occasional new product we're evaluating like Affine. The proprietary tools sit at the edges. The core of what we run -- the operating systems, the web applications, the databases, the file sharing, the communications, the configuration management, the build pipelines -- is open source or custom and in version control. Any sufficiently determined engineer could pick up our repositories tomorrow and rebuild our entire operation on different cloud providers, on bare metal, or in their basement.
That's the thing SaaS cannot offer. In a SaaS world, you don't own anything. You rent. The vendor decides what features exist, when they change, when they cost more, when they shut down. Your data is in their format, on their infrastructure, accessible on their terms. When they get acquired, your operations become a footnote in someone else's strategic plan.
The European Union has spent the last few years building out the legal and political case for digital sovereignty -- the principle that organizations and nations should not be structurally dependent on American hyperscalers for their critical systems. We agree, and we extend the principle to individual organizations: you should own your business systems, not just rent them.
This is harder than SaaS. It costs more in the short term. It requires actual expertise to do well. That's exactly where we come in. We help organizations run their own open source infrastructure -- web applications, file sharing, communications, configuration management -- so the data is theirs, the timeline is theirs, the decisions are theirs.
Control is the thing the cheap options cannot give you, no matter how much you spend with them. The SaaS vendor sells you access, not ownership. The cloud hyperscaler sells you a tenancy, not a home. Every dollar spent there buys convenience, not control -- and when you eventually need control, you discover you've been buying a thing that does not include it.
You don't own anything you can't rebuild.
Human-scale systems
Architectures that can be understood by the people running them. Stacks built from replaceable components rather than monolithic platforms. Configuration in version control so anyone with access can see what changed, why, and who did it. Documentation written for humans. AI agents working with explicit context, not guessing. The opposite of black-box magic. The opposite of "trust us."
We've spent twenty years building infrastructure this way. The specific tools we use -- Matrix instead of Slack or Teams, Nextcloud instead of Google Drive or Microsoft 365, Forgejo instead of GitHub, NixOS for declarative server configuration -- are all chosen because they can be understood and replaced. No single component is the system. The system is the architecture.
Federation and small, cooperating businesses
Our mission has been the same for over two decades: to make small and mid-sized businesses outperform big business on the things that matter. Not just compete. Better. The dominant economic narrative says that bigger is more efficient, that consolidation is progress, that the future belongs to the few players large enough to operate at planetary scale. We think that narrative is wrong, and we think it can be proven wrong one organization at a time.
Big businesses have scale, capital, and lobbying power. They will always win on those dimensions. What they cannot do is move quickly. They cannot serve a customer personally. They cannot make a judgment call that violates corporate policy because the customer's situation is unusual. They cannot stay close to the communities they operate in. They cannot resist the pressure to optimize shareholder returns at the expense of everything else.
Consider a story. In 2025, a developer in Morocco named Abdelkader Boudih watched AWS delete his ten-year-old account without warning. He had no recourse through normal support channels -- those had been automated away. Then a 20-year open source veteran inside AWS named Tarus Balog read Boudih's blog post, felt something, picked up the phone, escalated past every bureaucratic barrier, and got the CEO's attention. The account was restored. A formal "Correction of Error" process was launched to make sure it wouldn't happen again. Ten months later, AWS fired Tarus. His proudest accomplishment in four years at one of the largest companies on the planet -- saving one developer's data -- was not the kind of thing the system was set up to reward. Eventually, the system did what systems do: it optimized him away.
That's the dynamic in microcosm. A large organization briefly contained a human who could act like one, and the structure couldn't tolerate it. The "customer obsession" was real for one person and structurally impossible for the institution. This is what big businesses cannot do, even when they want to and even when they have the right people inside them. Small organizations, configured well, can.
That's the work we want to do, and the kind of organization we want to do it with. The small nonprofit running on infrastructure it owns. The regional manufacturer keeping its data sovereignty intact. The community media organization using federated tools to reach its audience without renting access from a platform. The mission-driven business that refuses to grow into the thing it used to compete with -- that, under sufficient venture pressure, has not quietly removed "Don't be evil" from its founding documents. The independent shop that has stayed independent long enough to remember why that mattered. The B Corporation committed to demonstrating that profitability and public benefit are not opposed. The Zebra company explicitly rejecting the unicorn-or-bust model that has captured so much of the startup world. These organizations exist. They are increasingly looking for partners who share their values. We want to be one of those partners.
We are not building a market of underdogs. We are building a different shape of economy.
Scale this up -- across industries, across regions, across the global open source movement -- and what you get is not a market of underdogs. It is a different shape of economy. Small, loose federations of small companies working cooperatively. Wealth distributed across many owners rather than concentrated in a few billionaires. Power distributed across many decisions rather than centralized in a few corporate boards. Resource use distributed across many communities rather than concentrated in a few data centers drinking the local aquifers. Knowledge and labor staying in the places they came from, integrated with the actual lives of actual people, rather than abstracted away into the spreadsheets of a holding company in another time zone.
The current direction of the economy is the opposite -- concentration, consolidation, extraction, scale. The small businesses worth working with are sailing against a strong current, and we are one of them. The destination of that current, if we let it carry us, is a world in which the work of small organizations is impossible, in which entrepreneurship means "build something a billionaire will buy," in which the local economy is something we read about in history books.
The cooperative web of independent shops, open source projects, and federated services isn't nostalgia. It's an active answer to the question of what to do when the alternative is monopoly. It's also, we believe, the better answer for almost everyone who isn't already a billionaire.
Stewardship of the communities we belong to
We succeed because we're members of many communities: the Drupal community, the open source community, Seattle, the Pacific Northwest, the Earth. Our work is meant to leave those communities better off. We contribute back through services, volunteering, and donations. We support efforts like Acquia's Fair Trade Initiative that try to make the funding of the commons as structural as our dependence on it.
The commons does not maintain itself.
Why this matters now
The list of plausible catastrophes coming for organizations that depend on consolidated platforms is getting longer, not shorter.
The SaaS product your business runs on gets acquired by a company with different priorities, and you get ninety days' notice to move your operations somewhere else. An AI system with production access makes a confident mistake and wipes out data that took your team years to accumulate. The cloud provider that hosts everything you do has a bad day, and now you do too. The AI-assisted code your team is shipping turns out to contain vulnerabilities at ten times the historical rate -- and the same AI is helping attackers find those vulnerabilities faster than your team can patch them. Climate change is altering the physical infrastructure that data centers depend upon. The political and economic environments around all of it are volatile.
None of these risks is solved by routing more of your operations through a smaller number of large platforms. All of them are made worse by it.
The response to compounding risk is not consolidation. It's diversification. Redundancy. Resilience built into the architecture rather than promised by a vendor's SLA.
The cheap option looks cheap until the risk it carries shows up on your doorstep. And the risk it carries is not a hypothetical you can decide to accept later. It is being assumed every day you stay on a platform you don't control, ripening quietly, waiting for the conditions that turn risk into incident.
The people who understand this -- who have been saying it for years, often without being heard -- are the people who maintain open source infrastructure, who run small independent shops, who insist on owning their stack, who still know how to do things from first principles when the abstractions fail.
This is the work we want to do, with organizations that understand why it matters.
Who this is for
If you've read this far, you're probably one of two kinds of people.
Either you already suspected most of this and you were looking for someone to say it out loud -- in which case, welcome, you're not alone, and there are more of us than the prevailing narratives suggest. Find your federation. Join it. Build something.
Or you're skeptical, and you've been reading to figure out whether I'm crank or signal -- in which case, fair enough, I'd invite you to watch what happens over the next few years. The leviathans will continue to grow. The acquisitions will continue. The free tiers will continue to vanish. The AI mistakes will accumulate. The Cloudflare outages will recur, each one a little more consequential than the last.
Better early and unfashionable than late and vindicated.
Freelock is a company that's publicly, explicitly, unapologetically on the people side of the argument. We build with open source because open source serves humans rather than extracts from them. We own our stack and rent our metal because owning your infrastructure is the only way to actually own it. We work cooperatively with other small companies because the federation is stronger than any of its members alone. We believe the consolidation of technology under a small number of enormous companies is a problem that needs to be reversed -- and the only way to reverse it is to keep building alternatives, keep funding the commons, and keep refusing the story that this is the only way things can be.
If you want help building something on those terms, we're here. If you want to talk about what these terms mean for your own organization, we're here for that too. And if you just want to know that there are companies taking this position publicly so you feel a little less crazy for taking it yourself -- well, here we are.
References
- Our Philosophy — the longer version of our mission and principles
- Freelock AI use policy
- Open Source Solutions for Small Business Problems — John Locke, 2004
- Black Vodka, a Finnish Sauna, and a Database — the Monty Widenius visit, 2009
On industry consolidation
- Killed by Google — the community-maintained Google graveyard
- Heroku's free tier shutdown — Salesforce ending an era
- Adobe's failed Figma acquisition
- Cloudflare market share statistics
On AI
- Adam Conover on the bipartisan AI backlash
- Magnifica Humanitas — Pope Leo XIV's encyclical on safeguarding human dignity in the time of artificial intelligence, May 2026
- CorridorKey — Niko Pueringer's open source neural green screen tool
- Veracode: AI-generated code security risks
- Apiiro: the 10× security spike in AI-assisted repositories
- AWS Fired the One Employee Who Gave a Damn — Abdelkader Boudih on Tarus Balog's firing
- Amazon Web Services — Four Years and Out — Tarus Balog's own account
On open source licensing and forks
- What it costs to run Drupal infrastructure — Dries Buytaert on the funding gap
- DrupalCon Chicago 2026 keynote
- Acquia Fair Trade Initiative
- MariaDB Business Source License
- HashiCorp adopts BSL — an era of open source might be ending
- OpenTofu — the community fork of Terraform under the Linux Foundation
- OpenBao — the community fork of HashiCorp Vault under the Linux Foundation
On the alternative shape of economy
- B Corporation — certified businesses balancing profit and public benefit
- Zebras Unite — the cooperative movement of zebra companies
Open source tools we use and recommend
- Matrix protocol — open, federated messaging
- Nextcloud — open file sharing and collaboration
- Forgejo — open source git hosting
- ActivityPub — the protocol behind federated social media
Add new comment