Sidekiq: Open Source to $7M+ a Year, Solo — the Finished Form of a 20-Year “OSS + Paid Edition” Strategy

Zero employees, zero sales staff, zero advertising, and $7M+ (about ¥1.05 billion) in annual revenue. Mike Perham, creator of Sidekiq, the Ruby background job processing library, has spent more than a decade proving out what is currently the most complete answer to the old problem of “making a living from open source.” His revenue as a solo developer ranks among the largest in the world, and yet the operation stays remarkably quiet throughout. Drawing on his 2017 Indie Hackers podcast appearance and a saas.group interview, we trace the internal numbers and the design behind them.

The Trajectory, on a Timeline

The starting point was a problem he saw firsthand with a client. The consulting client he was working for at the time was burning through a dozen-plus machines on background processing, and Perham thought, “we should be able to handle the same load on one machine”, and wrote Sidekiq. A concrete performance complaint came first. After launch, his own employer became Sidekiq’s biggest user, so daily work itself became live field testing. His side-project hours before going full-time ran roughly 20 hours a week, nights and weekends. The knowledge accumulated from maintaining Resque and Delayed Job, what actually hurts in this category, sits behind that roughly two-month build time.

A Structure Where “Free Does the Selling”

Sidekiq’s free edition is used worldwide as the de facto standard for job processing in Rails apps. Developers build trust on the free edition, and only the companies whose reliability and feature needs grow as their business scales go on to buy the Pro/Enterprise editions (batch processing, encryption, priority support, and more). In fact, 90% of new customers are inbound. There’s no sales, no marketing, no advertising at all. The spread of the free edition itself does the selling, which is exactly why one person can run a billion-yen-a-year business.

What matters is where the line between free and paid gets drawn. Perham drew it precisely: “free is plenty for hobby and small-scale use; once you’re seriously using it in a business, you’ll want the paid features.” Hold back too much from the free edition and it never spreads. Make the paid edition too weak and it never earns. The fact that 5-10 existing users converted to Pro the instant it launched, just six months after release, shows this line was drawn correctly from the very start. Designing this boundary is the actual substance of commercializing OSS.

No Discounts, No Hires, No Exceptions

Running roughly 800 customers (as of 2017) alone is only possible through a thorough elimination of exceptions. He never discounts, for anyone, under any circumstance. A uniform price means the entire process of negotiation simply disappears, eliminating sales overhead. Payments run almost entirely on Stripe, with invoicing limited to the very largest customers. Day-to-day support runs to about six emails, and he limits proactive customer outreach to once a quarter. Even incorporating the business was deferred. He ran as a sole proprietor until revenue exceeded a salary. In his own words: “you don’t need to worry about paperwork. The hard part is getting the money in the door.”

There’s a philosophy behind the pricing too. Pro at $1,000/year and Enterprise at $2,000/year are dramatically cheap by enterprise-software standards. Perham, recognizing that “the developer market isn’t used to paying for tools,” deliberately started low. Given that his consulting client had been able to replace a dozen-plus machines with one, $1,000 a year is a sum a company can recoup instantly out of infrastructure savings alone, and it clears budget approval easily. This “too-cheap” pricing underpins the no-discount, no-negotiation structure, and it shows up in the 50-100% annual growth rate as of 2017.

And Perham has turned down acquisition offers, hired no one, and quietly raised prices while running the same business. If Transistor’s calm company is the team edition, Sidekiq is the extreme, one-person edition of “calm management.” Untouched by VC growth pressure, alongside the Ruby community for over a decade, nothing demonstrates more clearly that “the freedom not to grow bigger” can coexist with revenue at the very top of the range.

What Didn’t Work — The Inspeqtor Lesson

Perham has a failure, too. A monitoring tool called “Inspeqtor,” an expansion into a different category, never took off the way Sidekiq did and was eventually shelved. Even with the same developer applying the same “OSS + paid edition” model, if you can’t secure a position as essential ecosystem infrastructure, it doesn’t replicate. This failure proves, from the other direction, that beyond individual skill, Sidekiq’s success depends on the position it occupies: “a place nearly every app is forced to pass through,” namely Rails job processing.

There’s a structural risk in the same place. The business’s ceiling and lifespan are tied to the rise and fall of the Ruby ecosystem, and dependency on a single-language community can’t be diversified away. The road from $80K/month to $7M+/year also took nearly a decade. This business’s compounding comes not from speed but from not quitting.

Conditions and Limits for Replication

For Japanese OSS developers, this case boils down to a single transferable point: not donations, but a paid edition for companies is the widest road to making OSS a profession. Against the sustainability problems discussed around GitHub Sponsors and OSS burnout, Sidekiq shows a path where you build a reason for companies to buy, in the form of features.

But the preconditions are heavy. The OSS in question needs to be a standard, essential pathway within its ecosystem. You need a category where you can carve out features that only enterprise users want (encryption, batch processing, priority support). And you need a life setup that can survive the 18 months to going full-time on roughly 20 side-project hours a week. Pick the wrong category and you get Inspeqtor, the contrast between one person’s success and failure teaches that lesson more clearly than anything else could.

Related Reading

• Photopea — Another answer to running a free tool solo
• Transistor — The team edition of the “calm company”