Digital Ocean is on fire as one of the hottest AI cloud providers of the last 5 years that no one thinks about.

+ August was Coco’s 2nd best month in company history.

August was another great month for us at Coco. We didn’t beat July for our best month ever, but it was the 2nd best month.

The most important accomplishment of the month was completely migrating our backend infra (one of those tools was actually Digital Ocean). So a lot of today’s newsletter comes from first hand experience now using the platform.

We migrated our entire back end infrastructure (DB, Servers, Jobs processing) you name it.

It was a hard project that took all summer, but well worth it. Our Costs have decreased significantly, we’re handling scale so much more efficiently.

We’re planning to 7x our messaging volume in Q4 compared to last year and without this project I wouldn’t have been as confident we can handle the scale.

Now that the rebuilds complete I believe we can scale to handle a 10x YoY increase.

Note: As always, none of what follows is legal, tax, investing, financial, or any other sort of advice. And I was never here 😉.

DigitalOcean (The cheap AI back end servers) is making a bet nobody else in cloud is willing to make.

Stock's up big off the 2024 lows, still well off its 52-week high of $187.50. Rev growing 15%. Management just told investors margins get worse before they get better, on purpose. Meanwhile every major AI coding platform just got bought by someone with a trillion-dollar balance sheet.

And yet: DOCN's playing a game none of Google, Microsoft, or SpaceX can play. Be cheap. Be close enough. Win the customers too small for anyone else to want.

Link to DigitalOcean's FY2025 earnings.

NI +207%. FCF +76% all the numbers you want to see them crush it. Even more impressive considering how much they’re investing in growth they still printed that much cash and profits.

Founded:

Aha Moment:

Growth:

Model:

Where We Are Now:

Here's how we're turning DigitalOcean into the default cloud for everyone AWS was never built for. And make it worth $100B.

AWS's egress pricing has 5 layers most people never see. Base rate. Cross-AZ charges. NAT Gateway fees. Regional transfer. Service-specific exceptions. One real example: a $4,403 headline egress estimate turned into a $7,523 bill. That's a 71% surprise.

DOCN already doesn't charge most of these fees. It just never told anyone.

The move: build a live bill comparator. Customer pastes their AWS bill, sees the DOCN number next to it, hidden fees stripped out and labeled. Market this straight at finance teams, not devs. "The cloud bill your CFO signs without a follow-up question."

Takeaway: Being cheap is a feature. Being predictable is the pitch.

DOCN already has a CLI (doctl), an MCP server, and a set of AI dev templates. None of them cover GPU Droplets or the GenAI platform. That's the gap.

The move: close it, then stop shipping three disconnected tools. Brand it as one thing, an AI-integrated CLI, built for devs who live in their terminal, not devs who want a drag-and-drop UI. That's Replit and Lovable's audience, not DOCN's. Let them fight over the UI crowd.

Takeaway: The dev who never leaves the terminal is the dev DOCN already owns. Arm them.

Every major AI coding platform got bought in the last year. Cursor, by SpaceX. Windsurf, split between Google and Cognition. Replit and Lovable, both locked into Google Cloud. Bolt.new, now on Azure. That door is closed.

The layer underneath isn't. Dify, Flowise, CrewAI, LangGraph, AutoGen, and LlamaIndex aren't 1-click apps anywhere on DOCN's marketplace. Nobody owns this layer because it's open source, not a company a hyperscaler wants to buy.

The move: ship official 1-click templates for all of them, pre-wired to a database and an inference endpoint. Get listed in their own docs, next to Pinecone and Weaviate. Fund the maintainers directly instead of trying to out-bid Google for a platform deal DOCN was never going to win.

Takeaway: You can't buy the coding platforms anymore. You can own what they're all built on.

The greatest move DigitalOcean can make is becoming 80% of the hyperscalers for 50% of the cost, then riding a market growing 20%+ a year for the next 5 to 10 years. As long as you're on that ride, you're making money. DigitalOcean can keep leapfrogging everyone else because it's the cheap option that's similar enough that developers keep choosing it. As the market explodes, its share just grows. There are very few markets like this one. Better to be in it than not.

Here's what worries me about that thesis, and it's the part the plays above don't fully answer: cheap-but-similar is exactly the position that gets commoditized the moment a bigger player decides to compete on price too. AWS doesn't need DOCN's margins to survive a price war. DOCN does. The market isn't pricing that risk in right now because DOCN's stock story is "AI upside," not "AI price war."

My real prediction: DOCN doesn't become the next hyperscaler and it doesn't need to. It becomes the best-run version of the discount option, and that's worth a lot more than the market is paying today, but a lot less than the $300B some of this math implies.

So here's the question worth sitting with. If being cheap is DOCN's whole edge, what happens the day a hyperscaler decides being cheap is worth doing too?