Fastly agreed to acquire Tracelight for $195 million in cash, the companies announced Monday. Tracelight, founded in Seattle in 2022 by two former AWS engineers, builds distributed tracing that stitches together the segments of a request that conventional observability tools treat separately: the CDN edge, any compute running there, the path back to origin, and the application itself. Fastly plans to fold the technology into its Observability product and offer it as a differentiated capability against Cloudflare and Akamai. Tracelight's roughly 40 employees join Fastly, and its standalone product will continue for existing customers through at least 2028.
The gap Tracelight fills
Application performance monitoring tools instrument the application and see everything from the moment a request arrives at the origin. What happens before that, at the CDN, is a black box represented by a single latency number in a log file. For a request that spent 340 milliseconds getting to origin, the application trace shows nothing about whether that time went to DNS, TLS negotiation, a cache miss, an edge function, or a slow backhaul.
Tracelight propagates trace context through the CDN layer and emits spans for each stage, so a single trace covers the entire path. That required integrations with each CDN's logging and, in some cases, its compute runtime. The company built connectors for Fastly, Cloudflare, and Akamai, which is presumably part of what made it attractive and which raises an obvious question about whether the competitor connectors survive the acquisition.
Fastly's strategic position
Fastly has struggled commercially against Cloudflare, which has a far larger free tier funneling into paid conversion and a broader product surface. Fastly's traditional strength is with large media and commerce customers who value fine-grained cache control and its Varnish-derived configuration language, and that base has been slowly eroding. Revenue growth has been in the low teens while Cloudflare grows faster from a larger base.
The bet here is that observability is where large customers feel the most pain and where a genuinely better product can win a decision. That is plausible. A retail engineering leader deciding between CDNs cares about a lot of things and cares intensely about being able to explain a latency spike during a sale. Whether a $195 million acquisition moves that needle enough is the question shareholders will ask, given Fastly's market capitalization sits under $2 billion.
The story is rarely the launch. It is what breaks, what ships, and who owns the mess at 2 a.m.
What happens to the competitor integrations
Fastly's announcement says the standalone Tracelight product continues and does not commit to maintaining Cloudflare and Akamai connectors indefinitely. Two Tracelight customers who use it with Cloudflare told us they read that as a wind-down notice and have started evaluating alternatives. That is the standard fate of a neutral tool acquired by one of the parties it observed, and it happened to several monitoring companies before this one.
The alternative would be for Fastly to keep the connectors and sell observability as a cross-CDN product, which would preserve the customer base and would mean investing in a product that helps customers use competitors. Companies rarely choose that. Cloudflare has its own tracing improvements in progress and Akamai has a partnership with Datadog covering similar ground.
The OpenTelemetry angle
Tracelight's implementation emits standard OpenTelemetry spans, which means the data lands in whatever backend a customer already runs: Datadog, Grafana, Honeycomb, or a self-hosted collector. That design decision is the reason the product got adopted, because it did not require replacing anything, and it also means Tracelight was never capturing the valuable part of the observability spend.
For Fastly the strategic value is in the instrumentation being unique to its platform rather than in owning the storage and query layer. That is a defensible position and it is worth less than owning the whole stack. Fastly executives said in a briefing that they intend to keep the OpenTelemetry output and to add optional storage, which is the sensible answer and the one that tests whether customers want another observability backend.
Consolidation in edge platforms
This is the third notable acquisition in edge infrastructure this year. Akamai bought a security posture company in March. Cloudflare acquired a small database team in May. The pattern reflects a market where the core product, moving bytes closer to users, has been commoditized and where differentiation has to come from what runs at the edge and what you can see there.
For customers the practical effect is fewer independent vendors and more capability bundled into platform contracts. That is convenient and it increases switching costs, which is the whole point. The advice that keeps recurring from architects who have been through a CDN migration is to keep instrumentation vendor-neutral wherever possible, which in this case means preferring the OpenTelemetry output over any proprietary format Fastly eventually offers.
For Tracelight's team the outcome is a reasonable one. Forty people and a $195 million exit four years after founding is a good result, and joining a company with a CDN to instrument gives the technology a distribution path it would have taken years to build alone. The customers who adopted it because it was neutral bear the cost of that decision, which is how acquisitions in observability have gone for a decade and which is worth remembering the next time a neutral tool becomes indispensable to your architecture.
Skarvonix will keep following this beat with reporting grounded in how systems behave outside the launch keynote.
- Edge Computing
