New developer fees, retired endpoints, a clampdown on “intermediary” data routing, and a Strava-built AI assistant. A breakdown for the athletes and apps caught in the middle.
On June 1, 2026, Strava’s overhauled developer program went live.
The biggest change to how third-party apps can touch your data since the company’s controversial 2024 API rewrite. If you use any app that pulls in your Strava activities, or you build one, the new terms are worth understanding. Here’s what changed, why Strava says it’s doing it, and what it means downstream.
What changed
The headline items, in plain terms:
-
A developer fee. Strava is replacing its old free, tiered API access with a flat subscription for around $11.99/month
-
New access tiers. Developers are now sorted into “Standard” and “Extended Access” classifications, with self-serve upgrades for small apps and formal review for everyone bigger.
-
Retired endpoints. Some endpoints (particularly those exposing aggregate or other-athletes’ data, like club details) are being sunset.
-
An intermediary clampdown. Apps that route Strava data through third-party “intermediary” platforms are no longer supported → a direct shot at services that re-pipe Strava data elsewhere.
-
The standing AI rules. Strava’s 2024 ban on using its API data to train AI models stays, as do the limits on showing one athlete’s data to anyone but that athlete.
-
A 90-day grace period. Existing developers get a transition window before the new rules take effect.
And one addition rather than a restriction: Strava is rolling out its own MCP server (a read-only, subscriber-only connector, launched with Anthropic’s Claude) that lets athletes ask AI assistants natural-language questions about their own Strava data. It’s the company’s sanctioned path for “Strava data in AI,” arriving just as it closes the unofficial ones.
Why Strava says it’s doing it
The framing from CEO Michael Martin is blunt: this is about AI scraping. He’s described AI companies relentlessly crawling public sites to feed their training data, degrading performance, and trying to evade API terms to get at Strava’s data. He’s also said Strava has turned down licensing overtures from major AI labs, and the company has publicly called out Perplexity for allegedly masking its scraping through aggregator services.
There’s a commercial subtext, too. Strava filed confidentially for an IPO in early 2026 at a reported ~$2.2B valuation, with annual recurring revenue approaching half a billion dollars, the bulk of it subscriptions. Tightening control over a proprietary dataset and of course and turning developer access into a revenue line is a sensible story to tell public-market investors.
Whether you read all this as principled privacy protection or pre-IPO moat-building probably depends on which side of the API you sit on. Both can be true at once.
What it means for athletes
For most athletes, day-to-day life won’t change much. Strava has consistently argued that changes like these affect only a tiny fraction of apps. The friction shows up at the edges: third-party tools can generally only show you your own data rather than surface it freely to, say, a coach; some niche apps that leaned on retired endpoints or intermediary routing may break; and if you want to interrogate your Strava data with an AI assistant, the blessed route is now Strava’s own subscriber MCP rather than a third-party pipe.
What it means for developers - and the rest of us
This is the part worth sitting with. Strava is, famously, a platform whose dataset is almost entirely uploaded by its users. Most of it captured on someone else’s hardware: your Garmin, your Wahoo, your Coros. A company in that position fencing off third-party access is a notable move, and it won’t be the last. As training data becomes a strategic asset, expect more platforms to add tolls, retire open endpoints, and funnel access through channels they can meter and police. The open-API era that built the modern fitness-app ecosystem is quietly narrowing. For anyone building on top of these platforms, the lesson is old but newly urgent: an API you don’t own is a dependency, not a foundation.
Full disclosure, since we build in this space: at athletedata we connect to Strava among 15+ platforms, so this touches us too. Strava data will leave our MCP tier when the grace period closes in late August, while our core coaching keeps reading your Strava activities exactly as before. The reason it’s a minor edit rather than a crisis is that we connect to devices directly wherever we can. When your Garmin or Wahoo talks to us at the source, Strava’s API terms never enter the picture. That’s our bias as builders, and weeks like this one are why.
None of this is the end of Strava, or of third-party fitness apps. But June 1 is a clear marker of where things are heading → more walls, more tolls, more data kept close.
Worth watching how the rest of the industry answers over the next 90 days.