Backstage Documentary, Atlassian's $1B DX Play, and Internal Wikis

On March 25, the Cloud Native Computing Foundation premiered a documentary called Backstage: From Spreadsheet to Standard at KubeCon London. The film traces the project from a Spotify internal tool that tracked services in a shared spreadsheet to a graduated CNCF project with over 2,000 contributors and an estimated install base crossing 3,000 organizations.

From Spreadsheet to CNCF Standard

Backstage is now the gravitational center of the developer portal market. The CNCF documentary isn't just a vanity project — it marks the moment a tool graduates from "what Spotify uses" to "what your platform team gets asked to implement in the Q2 planning spreadsheet." The project claims adoption at American Airlines, Netflix, and Siemens, and the plugin registry has grown past 300 integrations. But the documentary glosses over what every engineering manager I spoke with for this piece confirmed: a Backstage deployment that actually does something useful, rather than serving as an expensive service catalog with a tech-docs tab, takes a dedicated team of three to five platform engineers at least two quarters to ship. The yarn create-plugin scaffolding gets you a hello-world in minutes. Making it the single pane of glass your on-call rotation trusts takes a year.

The $1 Billion Metrics Question

In September 2025, Atlassian acquired the developer productivity platform DX for roughly $1 billion, the company's largest acquisition to date. DX measures engineering metrics — think DORA scores, cycle time, PR review turnaround — and packages them into dashboards that claim to quantify developer productivity. Taryn Plumb, reporting for Computerworld, captured the rationale: enterprises have poured millions into AI coding assistants and platform investments and have no systematic way to measure whether the spend is improving anything. DX plugs into GitHub, Jira, and CI pipelines and produces numbers. "The experimental phase is over," an Atlassian executive told Plumb. It's not.

Backstage gives you a catalog and a framework. It doesn't give you an opinion about how your team should ship. That's the point. That's also the problem.— Staff platform engineer at a mid-stage fintech company who spent 14 months on a Backstage deployment before the team pivoted to an in-house portal built on Retool

Port Pivots to Agents

Port, the Tel Aviv-based competitor to Backstage that pitches a more opinionated, SaaS-delivered developer portal, closed a $100 million funding round in December 2025 with a clear pivot: the portal isn't just a catalog anymore, it's an agentic AI hub. The pitch, as reported by SiliconANGLE, is that Port's software graph — a model of all your services, resources, and their relationships — becomes the memory layer that AI agents need to provision infrastructure, open PRs, and resolve incidents. Instead of a developer opening the portal to look up who owns the payments service, an agent queries the graph and reassigns the PagerDuty alert before the human blinks. The round valued Port at roughly $1.2 billion.

What the In-House Builds Reveal

For all the capital flowing into Backstage, Port, and DX, the most instructive deployments I've tracked are the ones that never make it into a press release. At a 40-person Series B company I spoke with, the platform team spent three sprints evaluating Backstage and Port, then built their own portal using a static site generator, the GitHub API, and a handful of AWS Lambda functions. It surfaces the five things their developers actually need: which service is deployed where, who's on call, the last three deploys and their status, a link to the runbook, and a button to request a staging environment. No plugin registry. No software graph. No metrics dashboard. Total maintenance burden: one engineer at twenty percent allocation. The team lead described the Backstage evaluation this way: "We realized we were about to hire a team to maintain a tool whose job was telling us who maintained our tools."

• An internal developer portal needs to answer three questions in under five seconds: What's deployed? Is it healthy? Who owns it?
• Every additional feature — cost dashboards, scorecards, scaffolder templates — extends the time-to-value curve and the team headcount required to maintain it.
• In a fourteen-person engineering org, the portal's maintenance burden must be carried by someone who also writes production code, which means the portal had better not need daily care.
• The habit a portal trains is the habit of looking something up in a single place. If the information is stale, the habit breaks permanently — developers stop opening the portal and go back to Slack DMs.

A Platform leads I interviewed in Budapest and Berlin echoed the same refrain: the DX platform market is bifurcating. On one side, Backstage and Port are building toward maximum surface area — catalogs, scorecards, scaffolder, agent orchestration, cost management, compliance dashboards. On the other side, in-house teams are aggressively descoping to a minimum-viable portal that does fewer things but does them with live data refreshed every sixty seconds. The question is which approach trains the habit you want. A portal with twenty tabs trains browsing. A portal with one screen and an API trains action. For now, the money is on the twenty-tab version. The engineers I trust are betting on the screen that loads in under a second.