OpenAI Bought a German Cloud Startup So Its AI Can Keep Coding After Your Laptop Goes to Sleep
“OpenAI just bought the extension cord for its AI coding assistant, because apparently "persistent" is the one feature five million developers couldn't wait another quarter for.”
Sally's not done with you yet.
Drop a URL, screenshot, or file and Sally will give you the honest truth.
Sally's Take
Let's set the scene. OpenAI's Codex is on an absolute tear, boasting 5 million weekly active users and a 400-plus percent growth spike since early 2026. Knowledge workers are flooding in, making up a full 20% of users and growing at triple the developer rate. So naturally, the thing OpenAI needed most was a company from Kiel, Germany that used to be called Gitpod, quietly rebranded to Ona, and had already convinced major banks, pharma giants, and sovereign wealth funds to run AI agents through its cloud environments. This is not a flashy acquisition. This is OpenAI buying the plumbing, and they know it.
Here is where the cynicism kicks in. Ona's whole value proposition is that it gives AI agents a persistent, secure cloud environment to live in so they do not die the moment a developer closes their MacBook. That is a real problem worth solving. But you are telling me that OpenAI, a company reportedly valued in the hundreds of billions, with Sam Altman on speed dial at every major sovereign wealth fund, needed to acquire a 2020-vintage Kiel startup to figure out cloud orchestration? This is either a very smart shortcut or a very public confession that Codex was missing a load-bearing wall. Thibault Sottiaux, Codex's Core Products Lead, practically said the quiet part loud when he noted Ona will 'help us make Codex easier to deploy securely across production workflows.' Translation: it was not easy before.
And the acquisition spree context here is delicious. This is OpenAI's fourth known acquisition in 2026 alone, following Promptfoo for cybersecurity in March, Torch for healthcare at roughly $60 million in January, and the Mac AI interface play with Software Applications. OpenAI is not building an AI lab anymore. It is assembling an enterprise software conglomerate at speed, using acquisitions as a cheat code for capabilities it should probably have been growing in-house. The financial terms on Ona were undisclosed, which either means it was embarrassingly cheap or embarrassingly expensive. Neither answer is flattering.
Think your work can survive this?
Drop a URL, screenshot, or file and Sally will give you the honest truth.
What Actually Happened
- •On June 11, 2026, OpenAI announced it agreed to acquire Ona, a cloud infrastructure startup based in Kiel, Germany, formerly known as Gitpod and founded in 2020.
- •Ona builds secure, pre-configured cloud environments and orchestration tooling that give AI agents persistent access to tools, systems, and context for long-running tasks, the exact infrastructure Codex needs to run autonomously when a developer is not actively present.
- •Ona's CEO Johannes Landgraf and the full team will join OpenAI's Codex group once the deal closes, pending regulatory approval. Financial terms were not disclosed.
- •Codex currently has more than 5 million weekly active users, a 400-plus percent increase from early 2026, with knowledge workers now comprising roughly 20% of its user base and growing at triple the rate of developers.
- •Ona's enterprise agent usage grew 13-fold in 2026, with clients including major banks, pharmaceutical companies, and sovereign wealth funds, and backers including General Catalyst, Shopify CEO Tobi Lutke, and Datadog CEO Olivier Pomel.
Who Got Burned
Anthropic is the most obvious name on the burn list here. OpenAI just made a direct move to compete with Claude Cowork and every other agentic sandbox platform like E2B and Modal, and it did so by buying proven enterprise traction rather than building from scratch. Those competitors now face an OpenAI that has Ona's 13-fold enterprise growth, its banking and pharma client relationships, and its persistent cloud architecture folded directly into Codex. Ona's existing venture backers, including General Catalyst and Crane Venture Partners, almost certainly got a fine return, so they are not crying. But every developer-tools startup quietly positioning itself as the "cloud brain for AI agents" just watched their most defensible differentiator get absorbed into the world's most popular AI coding product. The category did not disappear. It just got eaten.
Silver Lining
Here is the genuinely good news, and it is real: AI agents that can run persistent, long-running tasks in secure cloud environments without dying on a developer's laptop are actually useful. Not hype-useful. Production-useful. The fact that Ona already had major banks, pharmaceutical companies, and sovereign wealth funds running on its infrastructure means this is not vaporware getting stapled onto Codex. It is proven enterprise-grade tooling. For the 5 million developers and knowledge workers using Codex today, the practical upshot is that their AI coding assistant is about to get dramatically more capable for complex, multi-step workflows. Johannes Landgraf and the Ona team built something real and it is going somewhere it can actually scale. That is a win, even if OpenAI's acquisition pace is starting to feel less like strategy and more like aggressive grocery shopping.
Your turn. Drop something.
Drop a URL, screenshot, or file and Sally will give you the honest truth.